But these deals are built on weak foundations: The MP Materials rare earths deal rests on an expansive interpretation of Defense Production Act authority that the company itself flagged as a material risk in its SEC filing.1 CHIPS was subjected to an appropriations schedule that created annual cliffs and forced the program office into complex financial engineering to manage risk across budget cycles.

Industrial policy is here to stay, but how we fund it needs to change. A bipartisan consensus has emerged that the US government must intervene in private markets and influence private sector activity to fulfill critical economic and national security needs. Whether those efforts succeed depends on whether companies can trust that the government’s funding decisions will be sustained through year three, year seven, year ten. But when legal authority is improvised, or funding is contingent on annual appropriations fights, companies price in uncertainty — particularly in today’s volatile political environment. In the short term, this means that a political risk premium is embedded in every government-backed industrial deal, making projects more expensive, timelines shorter, and investment more hesitant. In the long-term, the ability of the government to sustain a multi-decade strategy as a consistent player in industrial intervention is at stake.

CHIPS exposes the fragility of the appropriations cycle

The appropriations process typically operates on short time horizons. Most programs and agencies are funded on a yearly basis and have strict expenditure requirements, with funding expiring if not spent — the U.S. code even requires that appropriations acts are titled as “An Act making appropriations (here insert the object) for the year ending September 30 (here insert the calendar year)” (emphasis added). Congress has maintained its constitutional power over the purse through annual appropriations for years.

The CHIPS appropriations did not follow this default: over $50 billion was appropriated in a single act, covering five years of appropriations across the CHIPS manufacturing and research programs. These funds were what’s called “no-year money” — they never expired. Congress seemed to understand that for an urgent national and economic security concern, longer-term appropriations and flexibility in spending is paramount.

However, Congress still constrained these funds, making it harder for both the government and industry to rely on long-term funding. The $39 billion manufacturing incentives appropriation included a provision for an alternative “allocation authority” — essentially a backdoor process by which Congress could undo allocations made by the President. The CHIPS and Science Act required that the President submit “detailed account, program, and project allocations” for the program, but then allowed Congress to make alternative allocations in the yearly appropriations bill, which would override the allocation made by the President. Determining the right allocations, conducting due diligence, and negotiating awards was a complex effort; Congress could exercise the allocation authority and unwind it all. While Congress gave the CHIPS team, and the executive branch by extension, the charge to make long-term investments and the requisite funding to execute, in practice the program still had yearly cliff-edges. This influenced deal structures, requiring legal and financial engineering to manage appropriations risk, and still left private companies partly exposed.

The Secure Enclave program is emblematic of the fragility of CHIPS’s investment decisions. Secure Enclave is a $3.5 billion defense procurement program designed to “ensure a secure supply of microelectronics for defense requirements.”2 Public reporting suggests that the executive branch intended for Secure Enclave to be co-funded, with CHIPS providing roughly $1bn for the program, and DoD promising to cover the remaining ~$2.5 billion. Devoting even just $1 billion of a $39 billion commercially oriented appropriation to a defense program with limited commercial spillover was a hard call, but this allocation was deemed necessary under the President’s authority.

But Congress unwound the executive allocation plan by compelling Commerce to fund the full $3.5 billion out of CHIPS -- and in so doing upended the CHIPS program’s portfolio planning. The consequences were immediate and disruptive. CHIPS had structured agreements based on portfolio-level math (as described by Todd Fisher here), and the finite pool of funding was precisely allocated to achieve the statute’s goals. In a blink, nearly 10% of that funding pool was directed to Secure Enclave, and the office had to renegotiate agreements with companies to account for the smaller available funding pool. Recipients understandably viewed this as a breach of faith — commitments were being revised not because of any failing on the companies’ part, nor because of a reassessment of the program’s goals, but because Congress redirected funding away from the prioritizations made by the executive branch.

Even with these challenges, the CHIPS and Science Act might be the high water mark of appropriations flexibility. But it’s still totally insufficient for the scale of the problem, and rather than fix this, oversight concerns have pushed Congress and appropriators toward additional constraints.

The MP Materials deal: ambitious but fragile

Rare earth minerals are a chokepoint in geopolitical competition, prompting government intervention. China dominates their production, processing over 90% of rare earths globally. Those rare earths are critical to advanced technology ranging from smartphones to laser-guided missiles. While the US has produced rare earths at the Mountain Pass mine since the 1950s, we’ve largely shipped them to China for processing. China’s recent retaliatory restrictions on imports and exports of rare earths led the Trump administration to negotiate an agreement with MP Materials (the company that owns Mountain Pass) to shore up this vulnerability.

The deal is an escalation in ambition. Unlike previous support for MP Materials, which came in the form of limited grants for specific technological innovations, the Trump administration’s deal attempts to address the supply chain problem by employing a broader array of tools, combining supply-side assistance, such as loans and equity investment, with demand-side guarantees in the form of a price floor and guaranteed purchases. Previous support was ineffective in part because it failed to establish a viable US customer base, prompting MP to continue exporting much of its output to China for processing. The new package aims to change that.

However, the deal has fiscal exposure. The Department of Defense committed to a ten-year price floor for neodymium-praseodymium (“NdPr”) oxide, a rare earth alloy produced at Mountain Pass, compensating for any shortfall if market prices decline. The size of the contingent liability is uncertain because the price floor is indexed to the Asian Metal Market price, largely driven by the Chinese market. Should China increase supply to depress the price (particularly if the goal is to deter Western capacity buildouts rather than maximize near-term revenue), US price-floor payments would increase. Our fiscal exposure is linked inversely to Chinese strategic actions.

By any reasonable standard, MP Materials is the most comprehensive industrial policy package the US government has assembled for a single company in decades. DoD also agreed to purchase all output from MP’s new magnet manufacturing facility at production cost plus a guaranteed profit margin, adjusted for inflation — that means MP Materials, which previously focused primarily on minerals, is now poised to be a vertically-integrated national champion of rare earth magnet production.

But the deal assumes sustained funding. The price floor payments, guaranteed offtake, profit guarantee, and the $350 million in preferred stock all depend on ongoing congressional appropriations and Department of Defense budget allocations.

Moreover, the deal relies on an unconventional interpretation of the Defense Production Act. Title III of the DPA, which governs financial incentives, has historically been used for targeted investments in specific industrial capabilities, backing purchase commitments, cost-sharing agreements, or grants tied to discrete production goals. The MP Materials package expands the toolkit to encompass a ten-year price floor, a production-cost-plus offtake guarantee, preferred equity, and warrants convertible to a 15% ownership stake in the company. No prior Title III transaction has combined this range of instruments on this scale or over this horizon — MP’s own lawyers flagged the “unconventional use” language in the firm’s SEC 8-K filing:

“given the unconventional use of DPA Title III authority, the need for the Department of Defense to secure additional funds in the future in order to meet its obligations in these Transaction Documents, as well as the heightened sensitivity and complexity of contracting with a government entity, particularly in a high profile industry implicating national security, there can be no assurances that the authorization of and continued support for the Transactions will not be modified, challenged or impaired in the future, which could have a material adverse effect on our business, prospects, financial condition and results of operations.”

The filing goes on to note that the uncertainty may be compounded by variable interpretation of statute, “future changes in the federal administration,” and the unpredictability of appropriations.

Because of the lack of long-term appropriations, a substantial portion of the package remains unfunded and without binding contractual obligation and legal improvisation creates risk.

The political risk to MP Materials is not confined to the deal itself. Because the authority underlying the package is contested, any other high-profile controversy over the DPA can splash onto it. The clearest recent example: the Trump administration reportedly considered using the Defense Production Act to “force [Anthropic] to tailor its model to the military’s needs,” which would have been an unprecedented intervention in a private company. The compulsion never came to pass, but its mere possibility stoked concern around the DPA at exactly the moment the law is working its way through reauthorization. A defensible deal like MP Materials could get swept up in congressional backlash triggered by an unfulfilled threat simply because both rest on the same legal foundation.

Toward a more durable industrial policy framework

Addressing these vulnerabilities to industrial policy requires changes to legal authority, appropriations, and where possible, institutional structure. These reforms fall into two tiers: changes Congress can make in the current session by amending existing authorities and appropriations practices, and the establishment of a standing institution that would create durable infrastructure for industrial policy across administrations.

1. Define the policy toolkit in statute

   To correct for the need to creatively interpret statute, Congress must explicitly authorize agencies conducting industrial policies with the full toolkit to succeed. Purchase guarantees, advance market commitments, price stabilizing mechanisms, options, futures, and yes, even equity investments, should be made available to government agencies. Bipartisan, bicameral bills, such as the SECURE Minerals Act and the Industrial Finance Corporation Act, have been introduced with robust, enumerated toolkits. These bills, or their core provisions, should move as standalone legislation or as riders on the next NDAA (the path of least resistance for defense-adjacent industrial authorities), or as part of the reauthorizations of the Export Import Bank or DPA.

   While Other Transaction Authority (OTA) can provide sufficient flexibility for many contractual arrangements and fulfill these same needs, it fails to address the underlying problem. Relying on a single, undefined authority invites political risk, provides no clear guidance to agencies, and ensures that every transaction carries legal risk factored into private sector pricing from inception. OTA should be normalized as a problem-solving tool for cases requiring creative structuring, rather than a substitute for explicit authority.

2. Make appropriations available for the long term

   Beyond enumerating the tools available to agencies, Congress must also make appropriations available for long-term commitments. One approach is to make industrial policy appropriations no-year and available without limitation like the alternative allocations in the CHIPS and Science Act. Concretely, this means a single lump-sum appropriation. Congress can require the executing office to establish milestones to drive performance and outcomes; if a recipient fails to deliver, clawback provisions could protect the appropriated funds from being wasted. While determining which violations should affect funding is complex, a transparent milestone system forces these decisions to be resolved in advance.

   Congress would have justifiable concerns with this level of latitude. Appropriations are a primary oversight mechanism and broad spending discretion raises risks that funds will be spent in misguided or corrupt ways. Yet successful industrial policy requires flexibility, judgment, and continuous resources to increase private sector investment at the scale we need — that might include extraordinary interventions like MP Materials. While the MP Materials deal has its flaws, industrial policy practitioners need comparable latitude to execute such deals when strategic conditions require them. And when a deal is made, the recipient should be able to count on its investment being backed by available funding.

   One way to reconcile this tension is to pair discretion with structural safeguards. First, deals like MP Materials should be transparent and competitive. One-off deals raise concerns about who benefits and why particular companies are singled out. Congress could require notification of such deals that includes an agency justification of necessity. This would create a procedural hurdle, but also a public record that enables oversight by Congress, inspectors general, and the press. Second, the default execution for some tools should be at the market level rather than for individual companies. Price floors, for example, could be available to all qualifying domestically produced NdPr oxide, up to a certain volume. One model is the Federal Reserve’s emergency lending facilities under 13(3), which Congress amended after 2008 to require “broad-based eligibility,” precisely to prevent single-firm bailouts. Agencies should approach industrial policy with the presumption that tools are designed to move a market, rather than subsidize a single company. Congress can preserve agency discretion to waive that presumption when justified, but agencies should bear the burden of justifying departure from the default.

3. Establish an independent body

   Over a longer horizon, the most ambitious reform would be to take another cue from the Federal Reserve, establishing institutional independence with bounded discretion. The Fed’s significant discretion is constrained by statute, authorized tools, and established norms; Congressional oversight is maintained through staggered term appointments, enabling periodic board renewal, and semiannual hearings that hold the Fed chair publicly accountable.
   
   A comparable institution for industrial policy, comprising an entity with enumerated authorities, dedicated funding, and regular reporting obligations to Congress, could mitigate executive-branch volatility and appropriations uncertainty. Recent Supreme Court decisions, notably Seila Law v. CFPB (2020), Collins v. Yellen (2021), and the ongoing uncertainty after SEC v. Jarkesy (2024), have complicated the establishment and maintenance of independent agencies, limiting Congress’s capacity to shield executive functions from presidential control. The legal terrain for independent agencies has narrowed, and any such entity would need to be designed within those constraints and with enumerated authorities tightly drawn to industrial policy functions. In the current political climate, norms alone are insufficient — the operational framework must be legislated.

Uncertainty compromises industrial policy

The United States has achieved bipartisan agreement on the need for industrial policy far more quickly than it has developed the institutional capacity to implement it. Uncertain appropriations and ad-hoc deals under tentative legal authorities leave our industrial policy efforts likely to fail. It is time to pass statutes to give practitioners the tools necessary to achieve their strategic objectives. Without those assurances, every government-backed deal will carry a political risk premium that competitors in other economies do not face, and we will simply not be able to generate the investment necessary to ensure our economic and national security.

We had anticipated some of the other legal and compliance challenges we’d face, NEPA chief among them, and planned and staffed up accordingly. At first, Davis-Bacon didn’t seem to demand the same level of attention. The law had been on the books since 1931, and DOL published the wage schedules that companies had to follow. We knew this could increase costs, and that the cost impact would then flow through to our commercial negotiations with our applicants. But we failed to anticipate that applying a 90-year-old law to a brand new program in a brand new industry would create implementation challenges far beyond what we had prepared for. Getting pulled aside by a lawyer to find out about retroactive application was my first indication of that.

Ultimately, Davis-Bacon implementation became a central issue for projects representing more than half of the program’s funds.1 This piece describes the range of challenges we encountered and offers recommendations for future implementers and policymakers.

What is Davis-Bacon?

The Davis-Bacon Act requires that workers on federally funded construction projects be paid no less than the “prevailing wage” for their trade in the local area, as determined by the Department of Labor. DOL publishes wage schedules covering four construction categories — building, residential, highway, and heavy — with rates set by locality, usually at the county or multi-county level. Each schedule lists worker classifications (carpenters, electricians, ironworkers, pipefitters, laborers, and many others) with their own required wages and fringe benefits.

As of 2018, DOL published more than 130,000 individual wage rates in total. Rates are determined through voluntary surveys of local wages. Since 2023, DOL has used a three-step process to determine these rates:

1. If a single wage rate is paid to a majority of workers in a classification, that becomes the prevailing wage.

2. If not, the rate paid to at least 30% of workers prevails.

3. If no rate applies to at least 30% workers, a weighted average is used.

Because union wages tend to be uniform and well-documented in survey data, this methodology frequently sets union-scale rates as the prevailing wage.

For each worker, contractors must first identify the applicable wage schedule for their county and construction type, and then match the worker’s actual tasks to the correct classification within that schedule. Pay is based on what the worker does rather than their job title; workers performing multiple types of work in a single week may require multiple classifications, each with its own wage rate and fringe benefit requirements.

Davis-Bacon was not part of the original CHIPS authorization, which passed in the 2020 National Defense Authorization Act without funding. It was added during the Senate markup of the United States Innovation and Competition Act in 2021, on a party-line vote with one Republican addition. When the CHIPS and Science Act was enacted with funding in August 2022, Davis-Bacon remained included.

Davis-Bacon protects construction wages, but adds costs for programs like CHIPS

The rationale for Davis-Bacon is straightforward: it’s a longstanding policy of the federal government to ensure that construction workers get paid a strong wage on federally supported projects. That’s a reasonable objective in its own right, independent of what CHIPS was trying to achieve. And there was some collateral benefit for CHIPS too: prevailing wage can help attract construction workers to the job (though usually by pulling workers away from other projects in the region).

That said, the policy rationale for Davis-Bacon is distinct from that of CHIPS, and compliance came at a cost. Those costs compound: first, some projects become more expensive to build; second, the government absorbs higher costs through its incentives; and third, higher incentives for individual projects means less overall investment gets unlocked.

Project-level costs

I don’t have program-wide data for the costs that Davis-Bacon added to CHIPS projects, and the impact varied significantly by locality and project type. As a ballpark, for some of our leading-edge fabs — which cost $20-25 billion to build — the Davis-Bacon premium was in the hundreds of millions. The premium was generally a smaller share of total capital expenditures for leading-edge fabs, since a larger share of overall investment goes to equipment rather than construction labor. For projects more labor-intensive relative to their total cost, such as smaller supply chain facilities, the share could be higher. But the cost impact depends on locality and project specifics. It also depends on labor market conditions: if construction demand pushes wages up, Davis-Bacon becomes less of a binding constraint.

Program-level costs

The core premise of CHIPS incentives was to close the cost gap between building in the United States and in Asia. To achieve that, we designed our NOFO to provide just enough incentive to make each project’s returns meet the company’s cost of capital. That means an increase in construction costs leaves the government with a choice whether to increase the subsidy to offset the additional cost. Absent a higher subsidy, the firm might limit their investment plans or even decide to pursue its project in another country.

For programs like CHIPS, where the government funds only a small share of total project costs, the impact of Davis-Bacon on program efficiency can be significant. Imagine a $100 project with a $10 incentive — if project costs rise by 1% to account for Davis-Bacon requirements, you need an $11 incentive to reach the same economic outcome. That’s a 1% increase in project costs but a 10% increase in program costs.

Impact on overall investment

Those per-project costs then limit the overall size of the portfolio. The promise of programs like CHIPS is that by providing a relatively small incentive, the government can catalyze large amounts of total investment. A $100 program averaging a 10% incentive will catalyze $1,000 in total investment: government grants cover 10% of costs, and a mix of private capital, tax credits, and state and local incentives cover the rest.

But the program’s impact is very sensitive to the program-wide funding ratio. If the same program needs to average an 11% incentive rather than 10% to compensate for a Davis-Bacon premium, the government grants are now covering a larger share of the individual projects, so the same $100 only catalyzes around $900. Applied to a multi-billion dollar program like CHIPS, the impact can scale quickly.

Subscribe now

Implementation challenges

Notwithstanding the added cost to projects, Davis-Bacon generated relatively little friction in the CHIPS program’s early months — far less than NEPA did. Companies understood it was in the law, incorporated the cost impact into their funding requests, and we accounted for it in our underwriting and negotiations. It was a cost, but appeared to be a manageable one, at least for our major applicants.

What created more friction in our negotiations were the policy and administrative challenges of implementing the prevailing wage requirement. While not every company faced all of these challenges — many negotiations proceeded without trouble around Davis-Bacon — the law’s requirements created significant complications for deals representing more than half the program’s funds. Programs like CHIPS don’t operate on averages. Every project had a national security rationale, and losing any one of them was a bad outcome we worked hard to avoid.

Four challenges proved hardest to navigate: retroactive pay, paying company staff, determining wage levels, and enforcing weekly pay schedules.

1. Retroactive pay

When federal funding arrives on a project already under construction, Davis-Bacon regulations require that recipients apply prevailing wages retroactively. This proved to be our most significant challenge.

Some context helps illustrate the implications. Starting in 2020, as Congress considered legislation to incentivize construction, firms such as TSMC, Intel (in Arizona, Ohio, Oregon, and New Mexico), Samsung, and others began announcing major domestic semiconductor investments. The NDAA authorized CHIPS without funding at the end of that year. The companies began breaking ground in 2021 and accelerated construction investment through and beyond the passage of CHIPS funding in July 2022. By the time we were negotiating term sheets with leading-edge applicants, several fabs had made meaningful construction progress. Because the early stages of construction are more labor-intensive than later stages (which focus on equipment installation), a significant share of spending on construction labor for these fabs had already gone out the door.

The reality of these ongoing construction projects informed how we designed the program in 2023. And, after some internal deliberations, we decided it was necessary to fund projects already under construction. The US government had given no indication that ongoing construction would be ineligible for CHIPS incentives, and both the Trump and Biden administrations had actively encouraged new construction. Moreover, we didn’t want to create perverse incentives, either for CHIPS investments or future programs: if companies learned that breaking ground before federal funding was awarded would disqualify them from receiving it, rational actors would stop building and wait for the check — exactly the opposite of the investment momentum industrial policy should try to generate. Significant optionality also remained even for projects well into construction, as it is common for semiconductor companies to pause after completing a fab’s shell before deciding to spend $10 billion or more on specialized equipment; funding these projects would help ensure their completion.

Most importantly, the companies desperately wanted funding for ongoing construction, and their boards expected it. This gave us leverage in our negotiations, and we settled on a strategy of funding existing projects as part of a broader negotiation to secure the maximum overall investment possible. This strategy served us remarkably well — we received leading-edge commitments that meaningfully exceeded market expectations.2

The requirement for retroactive compliance is an understandable anti-abuse provision — it aims to prevent recipients from purposefully attaching federal funds to the end of a project to avoid coverage on work already completed. That logic may make sense in many federal construction contexts. But for CHIPS, private companies had been making investments measured in the tens of billions, years before federal awards would be awarded, on terms that weren’t yet determined, and through a competitive process with no guaranteed outcome. It was hard to justify imposing retroactive Davis-Bacon compliance costs under those circumstances.

The financial and operational implications of retroactive application were significant. A leading-edge project might have 10,000–12,000 construction workers on site at peak, with a rotating workforce totaling perhaps 30,000 individuals over the project’s life. Working through 300-plus subcontractors across multiple tiers, retroactive application could require identifying wages paid to 20,000 workers who had already cycled off the project, determining what each worker should have been paid under Davis-Bacon, and paying the difference — resulting in hundreds of millions of dollars in additional cost. Apprentices posed their own challenge: under Davis-Bacon, they may be paid less than journeyman rates (usually around 70%), but only if enrolled in apprenticeship programs registered with DOL. If pre-award contractors had used unregistered apprentices, retroactive application would entitle those workers to backpay at full journeyman rates.

For some of our applicants, the challenge of retroactive application was so significant that they would not sign an award until we were able to eliminate or mitigate the administrative and financial burden. But addressing these issues required working through DOL — which brought its own set of challenges, as I’ll describe below.

2. Applying Davis-Bacon to company employees

Davis-Bacon applies to “construction activities” regardless of whether they are performed by external contractors or a company’s own employees. This is, again, an understandable anti-evasion measure, but it created serious problems for at least one applicant in our pipeline. This smaller company makes chips critical to our automotive industry and defense industrial base, and was proposing a project to expand domestic production rather than rely on foundries in Taiwan. Unlike our greenfield mega-projects, this project would expand and modernize existing facilities. And rather than rely exclusively on outside contractors, the company planned to use a mix of contractors and its own workforce, an approach enabled by the fact that it had trained its fab technicians to both service equipment in operating fabs and perform various construction tasks — electrical work one day, plumbing the next, and specialized cleanroom installation the day after.

But applying Davis-Bacon to company employees rather than contractors proved to be a big hurdle. Davis-Bacon required tracking every hour each employee spent on covered construction activities — by trade classification, with a different prevailing wage applying to each — and paying a wage differential for that portion of their work as distinct from fab operations work or non-Davis-Bacon construction work. The company also relied heavily on profit-sharing (where a portion of employees’ pay was tied to the firm’s profits) and Davis-Bacon’s guaranteed wage floor was difficult to reconcile with a pay structure that was inherently variable. Moreover, Davis-Bacon has a statutory requirement to pay wages weekly, meaning the company would need to change its payroll systems for a portion of the pay for a portion of its workforce.

These hurdles were daunting, but the challenge wasn’t just operational; it was also cultural. Differential wages would create inequities within the workforce, with similarly situated employees getting paid different amounts, risking internal grievance. This problem was compounded by the retroactivity issue described above — raising the prospect of a “Davis-Bacon Christmas” where some employees received checks and others didn’t for reasons the company’s management couldn’t defend.

After several months working through these issues, the company ultimately decided it couldn’t proceed with the award as originally scoped. We then spent another few months restructuring the project around work that would be performed entirely by outside contractors in an effort to salvage the key national security benefits of the deal. By the time we had a workable path, market conditions had shifted, management had changed, and the deal was lost. The Davis-Bacon implementation challenges had probably cost us half a year, and, as our CIO Todd Fisher used to say, time kills all deals.

Subscribe now

3. Determining applicable wage rates

A more mundane but still significant challenge was determining which wage rates applied to which work. The construction tasks involved in building and modernizing semiconductor fabs don’t always map cleanly onto DOL’s Davis-Bacon classifications, so applicants must go through a construction plan line-by-line to determine which rate applies to which activity. In traditional Davis-Bacon contexts this is less burdensome because contractors know the system and have processes in place. But semiconductor construction was a novel application, and all of our applicants — and most of their contractors — were navigating Davis-Bacon for the first time.

For large recipients, the administrative cost of this work was real but manageable relative to project scale: they could hire consultants, procure software systems, and build internal compliance capacity. For smaller companies, it was a more meaningful burden. To help, the CHIPS team ran dedicated Davis-Bacon workshops with applicants. These workshops typically entailed a few hours on a Zoom call with the company’s construction team, our investment team, and our internal Davis-Bacon team, working through large spreadsheets detailing planned construction activity to categorize that activity by Davis-Bacon classification.

Once awards were issued and construction was underway, a few specific issues made rate determination challenging. The county-by-county variation in DOL’s wage schedules is driven by the uneven survey data underlying them, which are based on voluntary surveys and can be sparse. As a result, the number of published classifications varies widely: some localities have detailed schedules covering dozens of trades; others have only a handful. When semiconductor construction involves work that doesn’t match any existing classification in a locality (which is commonly the case), the company must submit a “conformance” request. This process often results in required wage rates that seem disconnected from the local labor market. Moreover, because wage determination takes time (say, six to nine months), it can result in retroactive payment obligations and create uncertainty about the overall price tag for construction for projects already underway.

4. Weekly pay

The requirement for weekly pay can be a challenge too. Today, most employers pay on a biweekly basis, so weekly pay is an operational requirement that most companies outside the traditional federal construction ecosystem had never encountered. For our applicants, adapting payroll systems was a real but generally manageable compliance task. It created bigger challenges elsewhere: at DOE’s Loan Programs Office, several applicants negotiating large energy loans found the administrative challenge of weekly pay to be much more troublesome. More significantly, they had biweekly pay schedules written into existing union agreements that couldn’t be easily reopened. By my understanding, resolving this consumed months of senior-level attention, and in some cases the issues were never fully resolved.

On the whole, for our largest projects, none of these challenges stopped, or even meaningfully delayed, construction. Projects advanced, companies adapted, and compliance systems were built. The cost was measured in management bandwidth, negotiating friction, and program funds — not in abandoned megafabs. The one deal we lost was a smaller project, and Davis-Bacon was one of several contributing factors. My argument here is not that the law made CHIPS fail, but that it made CHIPS harder and more expensive than it needed to be. For smaller, more marginal projects, those costs can be decisive.

The challenge of split jurisdiction

Davis-Bacon’s requirements fall into a few categories: some are fixed in statute, some in DOL regulation, and some allow for more flexibility through waivers or other administrative action. That administrative flexibility was ultimately in the hands of DOL, which has final enforcement authority over Davis-Bacon. This meant substantial interaction between Commerce and DOL over the course of the program.

On some issues we got resolutions and on others we didn’t, in which case the company had to decide between accepting the additional burden or declining the award and investing elsewhere. In the case of retroactivity, we sought a programmatic waiver, but at the time DOL did not believe that a programmatic waiver was justified under their regulations and precedents. We looked for a path through on a case-by-case basis, with mixed results. Overall, we muddled through.

This was frustrating to me and my team, but I lay no blame at the feet of the public servants at DOL. It’s their job to administer Davis-Bacon broadly, which applies to thousands of programs and hundreds of billions of dollars in construction, and granting flexibility for one new program creates precedent. The tension is structural: denying flexibility had implications we’d have to live with at CHIPS; granting it had implications DOL would have to live with for Davis-Bacon for years to come. When two agencies with different mandates are both responsible for a program’s execution, the result is unlikely to be optimal for either’s policy objective.

Subscribe now

Recommendations

Based on CHIPS’s experience with Davis-Bacon, I offer four recommendations.

First, policymakers in Congress should carefully underwrite the fiscal costs of Davis-Bacon when weighing whether to apply it to new programs. For discretionary, partial-funding programs like CHIPS, the cost impact scales in ways that don’t arise in typical government programs: if federal funds cover only 10% of project costs, a 1% increase in total construction costs requires a 10% increase in the grant to maintain the same economic outcome for the project. That impact will vary depending on local labor markets, the makeup of construction costs, and macroeconomic conditions — but policymakers should go in with eyes open.

Second, in determining whether to include Davis-Bacon in a new program, Congress should recognize that implementation friction is cumulative. We weren’t just dealing with the challenge of Davis-Bacon — we were navigating NEPA, national security guardrails, federal interest requirements, the Paperwork Reduction Act, byzantine procurement rules, and a slew of other statutory dictates, all alongside the core commercial challenge of negotiating some of the largest industrial investments in American history. No single requirement broke the program. But the accumulation of legal and compliance requirements was a problem, and Davis-Bacon was among the heaviest weights in that stack. When designing industrial policy programs, Congress should consider the sum of compliance burdens, rather than assess each requirement in isolation.

Third, wherever Davis-Bacon applies to novel programs, implementing agencies need significantly more administrative flexibility than the current framework provides. Most urgent is the retroactivity requirement: Congress should either specify that Davis-Bacon compliance is prospective from the date of award, or create a clear statutory pathway for project-level or programmatic waivers. More broadly, the current split-jurisdiction model — in which DOL sets and enforces compliance standards for programs it isn’t responsible for delivering — creates structural misalignment that no amount of goodwill can fully overcome. Implementing agencies need a meaningful institutional role in adapting Davis-Bacon requirements to the programs they are accountable for executing.

Fourth, implementation teams cannot treat Davis-Bacon as an afterthought. It is a central implementation challenge that requires investment in staffing, internal capacity building, and applicant education from the very start of a program. We were underprepared — that cost us time and credibility with applicants, and it may have lost us at least one deal. Reducing reliance on Asia for chipmaking is a complicated undertaking; so is regulating construction wages. Putting the two together proved very challenging, and future implementers should plan accordingly.

I’ve spent most of my career in government — I’ve never had a private sector job aside from one summer in law school. At CHIPS I spent hundreds of hours negotiating with industry, and got used to hearing a litany of complaints from companies as part of those negotiations. But over time you learn what is posturing and what is real. The Davis-Bacon challenges described in this piece were legitimate, and both policymakers and implementers should take them seriously.

Macroscience

A better science is possible

By Andrew Gerard

Editor’s Note:

In August 2025, Secretary of Commerce Howard Lutnick cut funding to Natcast, a multi-billion dollar semiconductor R&D initiative enabled by the CHIPS and Science Act (also referred to as the CHIPS Act). Lutnick claimed that Natcast violated the law and accused the initiative of cronyism.

Taking a closer look at the program makes clear that these were misplaced accusations. The structure was neither radical nor corrupt, but rather a serious attempt to achieve the program’s outcomes under an ambitious mandate. It seemed like Natcast was on track to succeed — sunsetting it puts years of deliberate development to waste.

Today we bring you a piece that explains Natcast’s design from one of the leaders who set it up, Donna Dubinsky.

Although much attention has been focused on the $39B allocated by the CHIPS Act to build fabs,1 Congress also provided the Department of Commerce with $11B for research and development (R&D), the centerpiece of which was called the National Semiconductor Technology Center, or NSTC.2 The Act specifies that the NSTC should conduct research and prototyping of advanced semiconductor technology, grow the domestic semiconductor workforce, and establish an investment fund. Congress required that this effort be operated as a public private-sector consortium with participation from industry, academia, the Department of Defense, the Department of Energy, and the National Science Foundation.3

Implementing this vision raised difficult questions. What institutional structure could balance government oversight with private-sector commercial mindset and speed? How should research priorities be set, and by whom? Should the effort be short-term or long-term? Below, I describe the gaps the NSTC was designed to fill, the key decisions we made in standing it up, and the lessons future policymakers should draw from its creation — and its termination.

NSTC was established to fill gaps in US semiconductor R&D

While tens of billions of private dollars are spent on semiconductor research in the US each year, some clear gaps still remain.

• Shared research facilities: The cost of creating a modern semiconductor research fab or advanced packaging facility is in the multiple billions of dollars. Production fabs are optimized for volume, not experimental work. While big companies have dedicated research facilities, some research needs cannot be served in-house because of possible material contamination or disruption to the production workflow. Smaller companies and academics have constrained access to such facilities altogether. Other governments, including China, Japan, and the European Union, have made large public investments to establish shared research facilities which are used by companies from around the world. The US has nothing comparable, thus sending advanced research overseas.

• Pre-competitive research: Certain research will not yield products today but lays the groundwork for future development. Much of this work happens in universities or university-industry partnerships, while companies focus on near- to medium-term products. The NSTC would fund this longer-horizon research, bridging the gap between academic discovery and commercial application.

• Workforce development: Companies tend to invest limited capital in targeted programs to fill identifiable, near-term needs rather than uncertain, long-term needs. Government funds can be used to expand education to prepare next-generation workers for future jobs across the growing industry.

• High-risk investment: The cost and time from idea to market for new semiconductor technology has grown so large that most private investors cannot tolerate the risk. For example, the last major innovation of the semiconductor industry, the FinFET, took 17 years to mature from university research to production by Intel. Venture capital invested in chips has declined from almost 5% of total VC dollars in 2010 to just over 1% today. Government funds can stimulate private investment in high-potential American semiconductor start-ups.

The public-private consortium model was well suited to this effort: the NSTC vision required diverse private-sector skills (technical expertise, workforce training experience, venture investing knowledge), the capital to build shared facilities, and a direct collaboration to yield both national security and commercial benefits.

Implementation decisions

After the CHIPS Act passed, Secretary Gina Raimondo convened a team to evaluate how best to execute the NSTC. We gathered input from all constituents — the private and academic sectors (particularly the Industrial Advisory Committee formed under the CHIPS Act), the other agencies named in the Act, and the White House Office of Science and Technology Policy.

1. Grant program or an institution?

One foundational question was whether the NSTC should operate as a series of time-limited grant programs or as a lasting institution managing long-term assets.

Grant programs could be launched quickly using established government vehicles, but in deciding between a cluster of programs and a new institution, we needed to consider congressional intent. Most significantly, the Act mandated activities that required long-term investment at a large scale, particularly prototyping capabilities and advanced packaging facilities. These activities would consume most of the appropriations. While they could be implemented with the help of partners, they would need funding beyond the CHIPS appropriation. An institution that could recruit members, offer fee-based services, and build private-sector financial support was required.

2. What is the right institutional structure?

Having determined that the NSTC vision required an institution rather than a series of funding programs, we studied existing public-private partnerships and legal precedents to understand best practices. Two models were especially instructive. SEMATECH, a US semiconductor research consortium formed in the late 1980s, initially succeeded but collapsed after it stopped accepting government funds and became dominated by a few large firms. By contrast, imec, an independent nonprofit founded in Belgium in 1984, has thrived for over 40 years by maintaining independence while retaining approximately 20% public funding. The table below summarizes the models we examined.

Drawing on these lessons, we evaluated governance structures against the criteria our research had established: sector-wide independence, ability to attract senior talent, operational speed, capacity to attract private capital, and ability to create a strong government partnership. We concluded that we needed a new, focused nonprofit — purpose-built to attract senior talent, partner with the government, and earn industry credibility.

To comply with the Government Corporation Control Act (the GCCA), which prohibits the government from creating and managing a corporation without statutory authority, the Department selected experienced independent citizens through an open federal application to appoint a board. The initial board of trustees selected by the citizen committee included accomplished retired semiconductor executives, leading academics, and experienced corporate directors (I left the government and joined this board). The new board incorporated a nonprofit eventually named Natcast and hired an executive team who negotiated a contract with Commerce to operate the NSTC. This governance structure is not uncommon: SRI Quantum Economic Development Consortium, In-Q-Tel, and Natcast all operate through standard government contracts rather than government board control.

3. What is the right balance between government and private sector involvement?

Structuring the government-Natcast relationship was complicated. Too much government power risked politically driven decisions on programs and difficulty recruiting talent. Too much private-sector power risked national needs going unmet and domination by the largest companies.

The solution separated agenda-setting from execution. The government, Natcast staff, and a technical advisory board (comprised of leading technologists from industry and academia) would propose research topics; the government would approve topics and funding levels; Natcast would then run operations — awarding contracts and selecting recipients on purely technical and operational criteria such as feasibility, potential impact, and team experience, free from political interference or industry dominance. Natcast remained fully accountable through rigorous contractual obligations, national security compliance, and extensive reporting to Commerce. The same model applied to other programs such as workforce development and the investment fund; the government and Natcast together set strategic priorities, while Natcast selected and managed execution decisions.

4. When should we spend our appropriations?

We engaged a consulting firm to build a long-term financial model, proposing to use the appropriated funding — which fortunately did not expire — over 10 years, with preliminary budgets, membership structures, and program costs.

A strategic tension persisted. Many felt the goal was full self-sustainability by year 10. Others, myself included, believed ongoing government funding was critical — particularly for pre-competitive research that industry wouldn’t fund. When the entity is fully self-sustaining, there is no obligation to fulfill national interests, only corporate near-term objectives. The SEMATECH and imec case studies support this conclusion, with SEMATECH collapsing after declining government funding and imec succeeding over many years with ongoing support from the EC and the Flemish government. Since such continuing national support couldn’t be guaranteed, the model presumed self-sustainability — but one could return to Congress in five to six years and use successful outcomes to make the case for ongoing funding. In effect, the issue was deferred.

What we built

Natcast was incorporated in October 2023 and set about developing its own strategic plan, operating plan, and financial strategies. Although there were multiple major programs to create, Natcast made progress on all fronts:

• Early research programs were announced and some grants awarded; several other calls for proposals were ready to release.

• A Technical Advisory Board was established and a research agenda articulated.

• Partners were selected to build the prototyping center and the advanced packaging facility, and the EUV center was under contract, ready to start implementation.

• The Investment Fund was fully specified and had attracted substantial private interest.

• Workforce development programs were underway with academic institutions across the country.

• A variety of other programs were designed such as shared digital resources and a multi-project wafer capability

• Over 200 companies and institutions joined as dues-paying members.

Less than two years after creation, Natcast was executing well and meeting its goals.

What happened?

In late August 2025, Commerce announced its intention to discontinue Natcast as the NSTC operator. The contract was terminated on the pretext that Natcast’s creation violated the GCCA, despite extensive review on this question within Commerce’s legal department and a green light from the Department of Justice’s Office of Legal Counsel in the Biden Administration.

Commerce’s current plans are unclear. In September 2025, the Department announced a research grant program, but only one preliminary award has been made public. NIST suggested that the new approach “will mirror a more venture capital-style approach” and “strongly recommended that applicants include an approach to financial return on investment” for the government in their proposals such as granting the government equity, warrants, licensing, royalties or revenue sharing.

Beyond Natcast’s discontinuation (and the apparent termination of the NSTC itself), the Industrial Advisory Committee has been disbanded, the National Advanced Packaging Manufacturing Program is not active,4 the new semiconductor-focused Manufacturing USA Institute has been discontinued, and the Consortium Steering Committee has not met since the change of administration.5 As these activities are mandated by the CHIPS Act, it is not clear how Commerce intends to comply with the Act without substantially increasing staff — at odds with the administration’s push for smaller government. From the outside, the new CHIPS R&D vision appears more like a profit-driven investment program than a provider of core infrastructure benefiting all participants and prioritizing American national and economic security.

Lessons for future policymakers

Natcast arose out of careful study and deliberation with multiple stakeholders, fulfilling the ambitious mandate Congress passed in the CHIPS Act. Despite the substantial progress in realizing this vision, it was hastily abolished without a clear replacement plan.

While it might be impossible to legislate against this kind of reversal in the future, I believe that if Natcast had been six months further along, it is less likely that its work would have been halted because many more constituents would have been impacted. This underscores how critical it is to move quickly. The program was delayed, in part, by preferencing the incentives program for leadership attention, the need to work across multiple government agencies, and by the substantial delay needed to resolve legal issues to ensure compliance with the GCCA.

In thinking about future programs, policymakers should consider the following:

1. Match investment time frames with desired outcome time frames. To the extent that a policy has a long-term goal, it needs to be matched with a long-term investment structure. Had the R&D part of the CHIPS Act been viewed as a short-term need, say research programs and not facilities, it could have been clearly defined and easily implemented. But since the NSTC was envisioned as a long-term effort to become globally competitive, it needed the corresponding ability to execute over the long term. By contrast, the $39B fund for incentives was imagined and executed as a time-limited grant program.
   
   Two specific long-term aspects could have been enabled in the legislation. First, if the legislation had permitted the creation of a government corporation, that would have been a possible execution path. In that case, the effort would have been more insulated from political change because the governance structure could have had several trustees appointed by the current administration, enabling a new administration to impact the agenda without resorting to outright cancellation. Second, the question of sustainability could have been addressed through some notion of continued funding possibility, particularly for the research program, even if not an explicit commitment. A model that spelled out a funding renewal process (contingent upon program success) would have enabled planning that could presume ongoing government participation.

2. Designate clear responsibility for implementation. The CHIPS Act called for the Department of Commerce to lead implementation, giving a clear signal that commercial success was as important as national security. Congress provided separate funding to the Department of Defense to address national security needs. However, the Act also required Commerce to set up the NSTC “in collaboration with the Secretary of Defense” and it required participation from the Department of Energy and the National Science Foundation.
   
   While it is admirable to strive for agreement among multiple agencies, we found that the complexity of consulting so many agencies (as well as the White House) on so many programs caused unnecessary delay. For example, the Act could have said that Commerce would design the program in consultation with the other agencies, but not require active collaboration or participation.

   
   I estimate that, had we been able to create a government corporation and just consult with other agencies rather than fully including them in program design, the program could have been launched nine months to a year earlier — critical timing for an urgent program.

3. Establish a group of expert advisors. The CHIPS Act specified the creation of an Industrial Advisory Committee for the research effort. Appointing and convening this group was one of the earliest activities undertaken at Commerce. Although there were many other ways we received input, including RFIs, individual meetings, papers presented by constituents and industry colloquia, there is no doubt that the IAC was the most efficient means of getting high-quality input.

   
   An excellent group of technical advisors from industry, academia, and government was selected and their reviews and reports turned out to be invaluable for our work. Although bringing together a group of fiercely competitive companies risks inviting a battle of parochial needs, in this case, the individuals selected and the effective management of the IAC enabled a truly collaborative effort. Whether articulated in legislation or not, the thoughtful selection of key individuals from the private sector to advise the program design can surface relevant input efficiently.

Conclusion

The cancellation of Natcast’s contract dismantled a new institution poised to make a generational investment to address structural gaps in America’s semiconductor research ecosystem. The research agenda, the prototyping and advanced packaging facilities, the investment fund, the workforce pipeline, and the 200-member consortium cannot be replicated with a short-term grant program. What Commerce has proposed to replace the entire NSTC program is a grant program that demands equity stakes and revenue sharing from recipients, and is thus unlikely to attract broad industry participation. The damage will compound over time as researchers, start-ups, and industry partners redirect their efforts to better-supported ecosystems overseas.

The deeper lesson here is about policy continuity. Major semiconductor technology developments take decades and require collaboration between industry, government, and academia. China, Japan and the EC are executing semiconductor development plans well, with consistent funding and institutional support, often across leadership transitions. The United States will not succeed in this geopolitical competition if critical programs can be canceled every four years. The NSTC’s termination signals to industry, to allies, and to rival nations that American industrial policy commitments are provisional.

The creation and funding of the NSTC was a compelling, bipartisan effort to address clear gaps in our industrial ecosystem, and Natcast’s potential was promising. Future policymakers should strive to insulate such initiatives from political headwinds. NSTC and Natcast’s elimination is a missed opportunity for American leadership in the industries of the future and for our national security.

Donna Dubinsky has spent her career in Silicon Valley helping to advance technology. She is a serial entrepreneur and was CEO of Palm, Inc. and co-founder of Handspring and Numenta, a neuroscience-based AI company. She joined the Commerce Department in 2022, reporting to the Secretary of Commerce, to lead the Department’s implementation of the CHIPS Act. She left the government and subsequently served as an unpaid trustee at Natcast. These comments represent her personal opinions and not those of any organization or other individuals.

From early 2023 through early 2025, a small group of people — usually spanning investments, legal, ops, and the front office — was designing the detailed operations of the CHIPS awards process. This group needed to work two steps ahead of the rest of the team. It felt like paving the runway as the plane was starting to taxi.

We reflexively turned to familiar spreadsheets and tools to track our progress. In trying to make them work, we abandoned and resurrected the approach multiple times. This wasn’t a failure of discipline or buy-in. The problem was that we were asking a planning tool to do something it can’t: impose structure on a system that was still being designed.

What eventually worked was a deliberate meeting architecture, with recurring conversations at different frequencies and with distinct purposes. The meetings were the project management system and did what no single tracker could.

This is not an argument for more meetings — meetings are often wasteful, and most government programs have too many of them. Instead, I’m calling for meeting discipline that is calibrated to the kind of program you’re running and the uncertainty you’re operating under.

The CHIPS Program Office (CPO) ran 35 deals through a $39 billion investment pipeline in roughly two years, 20 of them to completion. The meeting structure I’ll describe here was a key part of how we pulled it off.

Detailed trackers can’t handle flux

Milestone trackers are accountability tools. They assume you can enumerate the steps of a process in advance, assign owners, and define completion. That approach was exactly right for hiring — the steps were known and the core need was to manage ownership and accountability across a large, fast-moving team. We also got some utility out of a higher-level tracker where we set one or two quarterly milestones for each strategic goal, but that was only useful for a couple of quarters before the urgency and unpredictability of deal-making made it feel beside the point.

Project management tools are not neutral systems. Each one encodes assumptions about what kind of work you’re doing. These tools are only useful when the process is stable, the sequence is known, and the main coordination problem is timely execution. None of those conditions were true for most of the challenges we faced in the first 18 months of CPO.

In that environment, a tracker was not just useless — it was actively misleading. Target dates were impossible to establish because we were still developing the process itself. Spreadsheets became an exercise in false precision, and updating them weekly was a time-suck that did little to advance outcomes. Worse, the ritual created the impression within the team that nobody really knew what was happening or how it all fit together.

In short, detailed trackers are not made for real-time coordination across a complex system that is still being designed. We eventually stopped pretending that a massive spreadsheet could do the work of orchestration. In a start-up environment like CHIPS, we need to allow for iteration, operational dynamism, and even a little bit of chaos. That required more meetings.

The four kinds of CHIPS meetings

Not all meetings serve the same purpose, and conflating them is how you end up with too many of the wrong kind. In retrospect I’d categorize what we did into four types of meetings, each with a particular function:

Management routines. These were the meetings that kept the team calibrated and aligned with the big picture. Our daily (then thrice-weekly, and then twice-weekly) leadership stand-up was a primary example — the meeting was brief, recurring, and low-stakes individually, but essential in aggregate. All-staff meetings served a different version of this function: communicating direction, celebrating progress, and ensuring that we were all working on the right things. When the team was small, we held all-staff meetings weekly. As it grew, we shifted to a monthly cadence. Management routine meetings are not the place for big decisions. They should be focused on prioritization, alignment, and building a sense of belonging across the team.

Troubleshooting meetings. These were called when something was blocking progress and we needed to get the right people in the room to fix it. Some blockers CHIPS had to handle included: a legal term in a contract negotiation that we couldn’t resolve at the deal-team level and needed executive judgment on; a tension between two internal teams about roles and responsibilities on a particular process step; a policy question about what was reasonable to ask of an applicant that required both program and legal sign-off. Troubleshooting meetings should be ad hoc, purpose-built, and end with a decision (or a path to one).

Governance meetings. These were the formal meeting series in which we debated deals and made decisions through formal, documented votes. CPO ran two versions of this meeting type: the Investment Committee (IC) meeting series, and the Transaction Review Committee (TRC) meeting series. For deals to move from one phase to the next, they needed to clear a structured set of decision gates in one or both of these meetings. Governance meetings gave clear direction to the deal team and finalized and formalized agency decision-making through documented votes.

Throughput meetings. These were recurring, structured meetings organized around moving specific work through a pipeline. Not “where are we on everything” — that’s too broad to be useful. Not “here’s a problem to solve” — that’s troubleshooting. Throughput meetings ask: what is in the pipeline right now, where is each item in the process, what does it need to advance, and who is responsible for making that happen before we meet again?

This last meeting type did the most work for us, and the outsized value surprised me. Throughput meetings effectively replaced the standard project management tools, transforming sequential updates into an effective and dynamic system for managing a rapidly evolving portfolio.

The throughput meeting in practice

By the time we were running dozens of deals in parallel — each touching 60 to 100 people across multiple process phases — we had built a stack of overlapping throughput meetings, each scoped to a specific audience or focusing on a specific slice of the pipeline:

• A weekly leadership-level deal review, run by the investments team off of a master deck that the team maintained and live-edited in the meeting. This was the only meeting where we looked at deals across all phases of the investment process simultaneously. We got direction on priority: if negotiations were stalling on one deal, we’d decide whether to escalate or pivot resources. This meeting let us signal urgency to all attendees and set the agenda for the rest of the program.

• Twice-weekly Investment Committee planning meetings, attended by the IC Executive Secretary, the investment operations team, Mike, Todd, myself, and Sara O’Rourke. These meetings looked at the month ahead and tried to anticipate when each deal would be ready to present to the Investment Committee, and for which decision-gate in the governance process. This required constant recalibration because a deal team’s ability to make deadlines often relied on getting new information from applicants or other external parties. This churn informed when we scheduled votes and when deals would be ready to move into the nitty-gritty processing steps that commenced after a deal was approved to proceed.

• Twice-weekly (then daily, then thrice-weekly, then weekly) transaction advisor meetings with NIST to ensure that we could solicit, select, onboard, and kick off due diligence of applicants that were advancing toward final award. This was a particularly complex and detailed part of our process, requiring coordination with internal and external stakeholders that made it tricky to calibrate the right cadence.

• Weekly internal diligence and final award term sheet meetings, focused on scoping the diligence and proposed final award terms for deals moving from preliminary memorandum of terms (PMT) into diligence and award phases. These meetings were originally troubleshooting sessions to align teams on what we’d ask of applicants, but after a couple of rounds they morphed into throughput meetings for tracking our progress.

• Twice-weekly award phase meetings internal to CPO, focused on tracking the back-end of deal-making, including drafting justification memos, ensuring conditions precedent to award were met, setting up awards in financial systems, and ensuring approval processes were on track with target award dates.

• Weekly award phase meetings with the National Institute of Standards and Technology (NIST) to coordinate across the institutional boundary that sat between us and the final obligating authority.1

In addition to these throughput meetings, we also spun up deal-specific processing meetings. There we’d review the list of conditions precedent to award and whether they were satisfied, enter and verify data into the grants system, and actually track the very final steps to closing and obligation. Each meeting had its own set of materials, scoped to its slice of the process. None of them tried to track everything, but together, they covered the whole pipeline. These meetings all evolved in either specificity or frequency.

Purposeful redundancy

There’s an obvious flaw in this architecture: it’s redundant. The same deal might appear in three different meetings in the same week, described to three different audiences. Some of the same attendees are in multiple meetings. This might sound like exactly the kind of meeting overhead that makes government slow.

But we found the redundancy to be necessary because the pipeline was moving too fast for any single meeting to carry the current state of reality. We would meet on Monday, align on milestones and next steps, and by Tuesday morning something would have happened — a company’s board had reconsidered a term, a new policy or legal interpretation had come down, or a deal team had learned something in a call that required resequencing. The next throughput meeting needed to account for the updates that happened during the week, not from last Monday’s notes. The only way for that to have made that work was to have people in the room who actually knew what had happened overnight.

Redundancy was critical for managing information streams of our volume. A handful of team members needed to regularly meet with the Secretary or other senior leaders and also attend operational meetings. We acted as translators who could connect what was being said in a planning session to a decision the Secretary had made two days earlier, or flag that a process the operations team was planning around had just been superseded by something new. Without that cross-level communication, the operations function would have been working from stale information, and the cost of that in a fast-moving pipeline is nontrivial.

The redundancy also maintained a near real-time shared understanding of the pipeline across a large team. The weekly review meeting’s decisions flowed down into the phase-specific meetings, and the phase-specific meetings surfaced issues that needed to come back up to deal review. The system worked because of the deliberate overlap and continuous flow of information.

A final irony

In fall 2024, as we were closing in on the end of the Biden administration and the most intense period of negotiation and drafting, we were asked to create a traditional milestone tracker. The request came from someone we couldn’t say no to: the Secretary herself. Having repeatedly seen the failures of false precision, Mike and I gave ourselves vague deadlines like “early November” and “late December.” But the Secretary wanted firm dates.

This time the trackers worked. By then we were managing a narrow set of late-stage steps for a defined set of deals that had already made it through merit review. The process was stable and specific enough to plan around. And the Secretary’s weekly review of that tracker then became a forcing function for the program, driving our next steps and informing every other meeting in the stack. The traditional tracker didn’t replace the meeting architecture, but was enabled by it.

That’s the thing I didn’t fully appreciate until later: the right tool isn’t the best tool in the abstract. It’s the tool that’s fit for the moment you’re in.

What this means for the next program

If you’re standing up a new program and seeking a framework: I’d recommend resisting the urge to immediately build the comprehensive tracker. Before you do that, consider your program’s maturity.

If your process is defined and you just need to establish accountability, a good tracker may serve you well. If you are still designing the process — if the flowchart doesn’t exist yet, or if you’re making foundational decisions about how the program works while also trying to move work through it — then you need a meeting cadence.

Before you build that cadence, consider your goals. Some meetings are about maintaining management routines that keep the team oriented and connected, separate from meetings dedicated to issue resolution. And if you’re running a transaction-based program, try building throughput processing meetings scoped to specific slices of the pipeline, attended by people who carry live intelligence across the levels of your organization.2

Accept that there will be redundancy. The cost of that inefficiency is lower than the cost of making decisions from stale information in a system that changes faster than any single document can track.

At the time, our work at CPO felt more like high-stakes improv than an actual functioning system. In retrospect, it was both, and I suspect that may just be a truism about what government execution at pace and scale demands.

Factory Settings

How to update the default settings of government, by former CHIPS Program Office leadership.

Briefing Book

The economic analysis policymakers get, from their expert former staffers

By Kevin Rinz

Editor’s note: Our first Factory Settings piece kicked up some debate about whether the CHIPS Incentives Program actually catalyzed semiconductor manufacturing construction investment. Detractors suggested that market forces were the true drivers, pointing out that construction planning began before CHIPS’s enactment and that construction spending was likely to occur regardless.

Analyzing the efficacy of public policy is important, and methodology matters. We tapped Skanda Amarnath, co-founder and Executive Director of Employ America, to write a comprehensive analysis of how manufacturing construction spend evolved over CHIPS’s legislative arc. The real options analysis Skanda uses here is a valuable lens for investigating industrial policy effects. His analysis makes clear that absent CHIPS, manufacturing spending would not have seen the same growth. We hope this piece sparks discussion about how to evaluate policy efficacy and invite your thoughts on the analysis.

The surge in US manufacturing construction spending since 2021 is one of the most prominent macroeconomic developments of the post-pandemic period. Construction of manufacturing facilities has risen sharply from a low base, contributing meaningfully to business fixed investment.1

The hockey-stick increase in spending is not in dispute. What remains less settled is how much of the surge is attributable to federal industrial policies — particularly the CHIPS and Science Act, signed into law in August 2022 — as opposed to other policies (like the Bipartisan Infrastructure Law or the Inflation Reduction Act) and market forces (like the advent of artificial intelligence models like ChatGPT) that would have driven manufacturing investment regardless. This question is key to both evaluating the return on public expenditure and understanding the structural forces shaping the US investment landscape going forward.

Skepticism about CHIPS’s effects on construction spending may stem from a January 2025 FEDS Note. The analysis uses project-level data from Dodge Data & Analytics to trace manufacturing construction from planning to construction starts.2 The central finding is that planning activity for new manufacturing facilities surged in 2021 — well before CHIPS was enacted — and that the mid-2022 peak in construction starts was driven by projects that firms began planning before the legislation became law. Based on that evidence, a natural inference is that the boom was already underway on its own market-driven logic, and that CHIPS may not have been so relevant to the historic rise in construction spending.

A more granular analysis of the planning data challenges that simplistic reading. The CHIPS Act had a long and public legislative history that made it possible to track the likelihood of passage well before its enactment; firms were likely planning their construction activity based on policy expectations even during the 2021 pre-passage surge the FEDS Note identifies. Breaking down the planning data by manufacturing subsector and census division makes apparent that a substantial share of the construction uptick was concentrated in the industries and geographies most relevant to CHIPS award funding, suggesting that the policy effects were more potent than a focus on the enactment date alone implies.

The conclusions we can and can’t draw from the Dodge data

The FEDS Note illustrates the manufacturing construction pipeline, analyzing Dodge data on more than 20,000 manufacturing construction projects dating back to 2019. The analysis tracks when projects entered the planning stage, how long they spent in planning, whether they were abandoned or revised in scope, and when they proceeded to a construction start. A planning start, as defined by the authors, is when a plan first emerges in the observed Dodge data, whereas a construction start is defined as “the commencement of physical work or the expectation of such work within the next few months.”

Planning starts for new manufacturing projects were very low in 2019 and 2020, mirroring subdued nonresidential structures investment in the pre-pandemic period. Activity surged beginning in 2021, peaked around the time of CHIPS and IRA enactment in August 2022, and has since trended down with some volatility.

Construction starts also surged (though at a lag behind planning activity) and peaked around mid-2022. That peak was largely driven by projects that had been in planning for less than one year. The planning pipeline has since grown substantially, but the share of newer plans (less than one year old) has declined since mid-2023. The authors note that it is unclear whether the aging pipeline is a positive or negative signal for future construction activity. If new plans emerge in anticipation of ripening policy conditions, it’s reasonable to expect a surge in new plans in the lead-up to policy enactment, and a decline in activity in the absence of further favorable policy shocks.

This analysis invites a range of plausible inferences, but the FEDS Note is careful in its conclusions, calling for continued monitoring rather than making strong claims about causation. But the timeline it documents — a planning surge that predates enactment — naturally invites a skeptical reading of policy effects. The key question is whether the enactment date is the correct timestamp for when policy began to shape firm behavior, or whether the relevant policy signal arrived earlier and through different channels.

Optionality is valuable

Planning activity could easily be construed as a response to enacted policies, but we can also evaluate it as anticipatory behavior. This shift rests on framing pre-construction activity in terms of option premium.3

Planning is a necessary but not sufficient condition for a project to be seen through to completion. Every stage of a project entails rational speculation about the likelihood that economic and policy conditions will ripen sufficiently to make the project viable. And while pre-construction activities can have substantial costs — including engineering and design work, permit acquisition, site preparation, and environmental assessments — they are often just a small percentage of the project total, which includes building, equipment installation and costs, and commissioning. As the Dodge data defines it, “planning is most typically associated with hiring an architect to create plans.” Given how little effort that entails, it is not a surprise that projects are more likely to be canceled in the planning stage than after commencing construction. In that context, planning is analogous to the cost of an option premium, while starting construction — presumably under ripe economic and policy conditions — is akin to exercising that option.

When the policy or economic backdrop is more uncertain and volatile, as was the case in the lead-up to CHIPS’s enactment, options have heightened value. Firms may be able to hold permits, maintain site options, and keep engineering work current while waiting for the policy backdrop to improve. To the extent that economic and policy uncertainty tied to the potential passage of CHIPS affected the economic viability of major projects, relatively low-cost planning was more valuable and served as a rational anticipatory step. The FEDS Note showed variable and sometimes long lags between planning and starts, consistent with how firms would seek to optimize when to exercise their options.

Options are particularly valuable in that they enable decisions in response to new information. The relatively modest expenditure on planning positioned firms opportunistically for a legislative breakthrough on CHIPS, enabling rapid scale-up in response to policy developments.

Per the Dodge data, construction starts appear to cluster around key policy developments. The FEDS Note’s finding that plans surged before CHIPS’s passage would be consistent with firms accumulating options when the CHIPS Act was a real possibility, but only starting projects after policy conditions, namely enactment, sufficiently ripened.

A critical feature of CHIPS’s implementation is that awardees were not required to start projects from scratch at the point of enactment. Firms that had already begun planning — or even preliminary site work or construction — were eligible for CHIPS awards. If firms anticipated these subsidies, it’s conceivable that the expectation could set investment in motion. Allowing subsidies for work that was already underway means it was even more likely that pre-enactment planning activity was a response to policy, rather than independent of it.

The CHIPS Act’s legislative arc

The strongest counterargument to timing-based skepticism is that the CHIPS Act did not arrive as a complete surprise. The legislation was the culmination of a multi-year, bipartisan effort that spanned two Congresses, multiple committees, multiple presidential administrations, and several distinct pieces of legislation. During that multi-year gestation, the probability of some form of federal semiconductor manufacturing subsidy moved from speculative to likely; the firms with the most to gain had strong incentives to position themselves in advance.

Tracing the legislative arc from inception to enactment helps clarify when the policy signal first became salient to firm investment decisions.

[NB: This granular breakdown of CHIPS’s legislative history illustrates the incremental developments firms may have responded to. A higher-level milestone summary follows.]

Phase 1: Origins and introduction (2019–2020)

The CHIPS and Science Act emerged from growing bipartisan concern about US competitiveness with China in critical technologies. In October 2019, Under Secretary of State Keith Krach presented a Global Economic Security Strategy to Senators Chuck Schumer and Todd Young, proposing a dramatic increase in government R&D investment with matching private-sector contributions.

Two parallel legislative tracks emerged in 2020. The first was the Endless Frontier Act, introduced on May 27, 2020, by Senators Schumer and Young alongside Representatives Ro Khanna and Mike Gallagher, which proposed a $100 billion expansion of the National Science Foundation to accelerate research in key technology areas. The second was the CHIPS for America Act, introduced on June 10, 2020, by Senators John Cornyn and Mark Warner in the Senate and by Representatives Michael McCaul and Doris Matsui in the House the following day. This bill called for $52 billion in federal incentives to restore US semiconductor manufacturing capacity, which had declined from 37% of global production in 1990 to roughly 12% by 2020. Alongside those legislative developments, the State Department successfully brokered TSMC’s $12 billion commitment to build a fabrication facility in Arizona, announced on May 15, 2020.

Phase 2: Authorization without appropriation (late 2020 – early 2021)

The CHIPS for America Act’s key provisions were incorporated into the National Defense Authorization Act (NDAA) for Fiscal Year 2021, which was signed into law on January 1, 2021. This authorized the creation of semiconductor incentive programs, a National Semiconductor Technology Center, and related R&D initiatives. However, the NDAA authorized these programs without appropriating any funding, meaning Congress would need to pass separate legislation to provide the $52 billion authorized.

At this stage, semiconductor firms could observe that a bipartisan consensus favored federal support for domestic chip manufacturing, but the probability of actual funding remained uncertain.

Phase 3: Senate action — the US Innovation and Competition Act (2021)

The 117th Congress saw rapid acceleration of legislation for semiconductor investment. The Endless Frontier Act was reintroduced as S. 1260 on April 20, 2021, and quickly advanced through the Senate Commerce Committee, passing by a bipartisan vote of 24–4 on May 12, 2021. Majority Leader Schumer then used the bill as a legislative vehicle for a much broader package: on May 18, 2021, he filed a substitute amendment that merged the Endless Frontier Act with contributions from five other Senate committees, including $52 billion in emergency appropriations to fund the CHIPS Act programs authorized in the FY2021 NDAA. The expanded package was renamed the United States Innovation and Competition Act (USICA). The Senate passed USICA on June 8, 2021 by a vote of 68–32, with strong bipartisan support. A 68-vote Senate majority for a bill containing $52 billion in semiconductor subsidies represented a qualitative shift in the probability of enactment.

The 2021 planning activity surge documented in the FEDS Note aligns with these dates. It’s quite plausible that firms saw the Senate vote as a substantive signal that federal subsidies were increasingly likely. Even if far from perfectly certain, subsidies may have been probable enough to justify the relatively low-cost step of initiating planning.

Phase 4: House action — the America COMPETES Act (2022)

Rather than taking up USICA directly, the House developed its own comprehensive competitiveness package. On January 25, 2022, House Democrats unveiled the America COMPETES Act of 2022, which also included $52 billion for CHIPS programs, but differed substantially from USICA in its science policy architecture, trade provisions, and inclusion of immigration reform measures. The House passed the America COMPETES Act on February 4, 2022, by a vote of 222–210, largely along party lines. On March 28, 2022, the Senate passed an amended version by substituting USICA text into H.R. 4521, formally creating the procedural conditions for a bicameral conference committee to reconcile the two bills.

As the House took up, and then passed, the America COMPETES Act, construction starts correspondingly surged as congressional action increased the likelihood of an eventual CHIPS passage.

Phase 5: Conference negotiations and passage (2022)

A formal conference committee convened in May 2022, but negotiations proved contentious. By early July, talks had stalled over disagreements on the structure of the proposed new NSF directorate, trade policy provisions, and other issues. Senate leaders considered advancing a narrower bill focused solely on semiconductor subsidies.

The semiconductor industry sounded concerns during this period, with companies explicitly conditioning major US investments on the passage of CHIPS funding:

• Intel was notably clear in its posture — in late June 2022, CEO Pat Gelsinger delayed the groundbreaking ceremony for Intel’s $20 billion Ohio fabrication complex due to congressional inaction. Speaking at the Aspen Ideas Festival on June 28, Gelsinger warned that without subsidies, Intel would redirect investment to Europe, where governments had moved faster to approve funding.

• GlobalWafers, a Taiwanese semiconductor wafer manufacturer, announced a $5 billion US factory plan on June 27, 2022, but made the investment explicitly contingent on CHIPS funding. Commerce Secretary Gina Raimondo amplified the urgency, stating that the GlobalWafers deal would collapse without congressional action before the August recess.

• TSMC board member and minister of Taiwan’s National Development Council Ming-Hsin Kung similarly signaled on June 28, 2022 that pace of construction and expansion of TSMC’s Arizona facilities depended on the passage of the CHIPS Act.

After a legislative breakthrough on July 19, 2022, the final package passed the Senate on July 27, 2022, by a vote of 64–33, and the House on July 28, 2022, by a vote of 243–187. President Biden signed the CHIPS and Science Act into law on August 9, 2022, appropriating $52.7 billion for semiconductor programs (including $39 billion for manufacturing incentives) and authorizing roughly $174 billion for science and technology investments.

Policy milestone summary

The following table summarizes the key milestones in the CHIPS and Science Act’s legislative journey, illustrating the mounting probability of enactment that firms could observe in real time.

The projects CHIPS supported: scale and location

Studying the scale and geographic distribution of CHIPS awards concretizes the connection between the policy arc and construction spending data.

As of February 2026, the CHIPS Program Office has issued preliminary or finalized awards to over 30 companies across over 45 project locations. The total direct federal funding committed is approximately $31 billion, with an additional $5.5 billion in government loans. As large as these awards were, they were dwarfed by the total private capital expenditure associated with these projects; estimates suggest nearly $400 billion in planned company investment across all CHIPS-supported sites. These capex figures span multi-year investment horizons; absent CHIPS, market forces may have necessitated some smaller portion of investment.

The largest awards and their geographic footprints

The distribution of CHIPS-associated investment is heavily concentrated in a small number of megaprojects. The five largest award groups — Intel, TSMC, Micron, Samsung, and GlobalFoundries — account for $27 billion of the $32.4 billion in direct federal awards and an estimated $304 billion of the $355 billion in total planned investment (86%).

The geographic distribution of CHIPS-associated capital expenditure is also regionally concentrated. While construction is largely still pending for projects in the Northeast, the bulk of the major projects were located in either Mountain census division states (Arizona, Idaho, Utah, New Mexico) or Texas.

Three census divisions — Mountain, West South Central, and Middle Atlantic — account for 80% of all CHIPS awards, and likely an even larger share of total firm investment. That said, most of the Middle Atlantic projects either have yet to begin or have only begun very recently. To investigate how CHIPS awards align with our observations of construction spending data, we should focus on construction spending that commenced over the relevant time period and in the regions where funded projects are located. $23.3 of the $32.4 billion in CHIPS awards were for projects where construction was confirmed to have started before 2025, and the Mountain and West South Central census divisions alone make up 65% of these CHIPS awards. If the CHIPS Act influenced construction spending, these two regions would likely stand out.

Triangulating policy effects

A cross-sectional analysis of the construction spending surge helps disentangle what’s driving the aggregate hockey-stick trend. This granular subsector data clarifies that the surge was a policy-driven and sector-specific boom rather than a market-driven cycle across all of manufacturing.

CHIPS-related subsector growth outpaced other industries

The construction spending increase is decisively concentrated in subsectors related to CHIPS efforts. When categorized by end-use subsector, Census Bureau data on manufacturing construction spending — reported as a seasonally adjusted annual rate (SAAR) — shows that the Computer/Electronic/Electrical category has driven most of the increase in total construction spending on manufacturing facilities.4

The Computer/Electronic/Electric category corresponds to NAICS 334, Computer and Electronic Product Manufacturing, and NAICS 335, Electrical Equipment, Appliance, and Component Manufacturing. NAICS 334 encompasses semiconductor fabrication and electronic component manufacturing — the direct target of CHIPS subsidies — and also spans photovoltaics and sensors that may have been supported by the Inflation Reduction Act (IRA), which was passed shortly after CHIPS; NAICS 335 captures batteries, electrical equipment, and related components that were intertwined with the tax credit provisions of the IRA.

Unfortunately for those of us interested in causal identification, the IRA was enacted just after the CHIPS Act, further muddying the analysis. But the concentration of nearly two-thirds of the incremental spending in the two subsectors most directly targeted by federal industrial policy is a meaningful signal, even if it fails to disentangle IRA and CHIPS-related support.

In the fourth quarter of 2019, construction spending in the Computer/Electronic/Electrical subsector ran at a SAAR of approximately $9.1 billion, representing 11.2% of total manufacturing construction; by the fourth quarter of 2022 — a few months after CHIPS enactment — this figure had risen to $63 billion (41.9% of the total). That proportion continued to climb, peaking in the fourth quarter of 2023 at $117.7 billion SAAR, or 52.4% of all manufacturing construction spending. As of the latest data release for October 2025, the three-month moving average stands at $97.6 billion SAAR (45.3%), marginally lower as many commenced construction projects tied to CHIPS reach completion. The increase in this single combined category accounts for roughly $89 billion of the $135 billion total increase in manufacturing construction SAAR since the fourth quarter of 2019 — approximately 66% of the overall gain.

If this surge were the result of generic post-COVID demand recovery, supply chain diversification, or commodity economics, we would expect to see broader distribution across food processing, chemicals, fabricated metals, and other subsectors. The Chemical subsector (NAICS 325) is the only other category that shows a substantial increase, rising from $32.7 billion SAAR in the fourth quarter of 2019 to $45.8 billion in the latest construction spending data release. This may be consistent with investment and production tied to the liquefied natural gas, pharmaceutical, or other specialty chemical industries, but its $13 billion increase is modest relative to the $89 billion gain in Computer/Electronic/Electrical. Food/Beverage/Tobacco, Transportation Equipment, and other categories have either been relatively flat or shown only moderate growth.

Manufacturing construction spending by subsector (SAAR, billions). The Computer/Electronic/Electrical category dominates the post-2021 increase, with policy milestone dates shown for reference.

To disentangle the investment effects of the CHIPS Act from the IRA, we have to rely on other datasets with greater sectoral granularity. These datasets tend to lag the Census Bureau’s timely monthly construction spending estimates by 2-3 years, but from what we can see, there is a prominent effect in the sectors most relevant to the CHIPS Act, and separable from the IRA. The latest data from the Census Bureau’s Annual Capital Expenditures Survey (ACES) — now within the Annual Integrated Economic Survey (AIES) — runs through calendar year 2022, just as the surge in manufacturing construction spending was getting going.

Based on the ACES data, total capital expenditures on computer and electronic product manufacturing structures (NAICS 334, more relevant to CHIPS) were estimated at $11.8 billion in 2022, of which $10.1 billion was estimated to be for semiconductor and other electronic component manufacturing. Compare this to the $1.3 billion of total capital expenditures allocated to structures for the manufacturing of electrical equipment, appliances, and components (NAICS 335, more relevant to the IRA).5 This suggests a roughly 90% to 10% split between the manufacturing of goods primarily pertinent to the CHIPS Act relative to those pertinent to the IRA (though this split is imperfect, since photovoltaic cell manufacturing falls under NAICS 334). From what we can glean from trade association data, semiconductor manufacturing employs ten times as many people as the photovoltaic cell manufacturing sector in the US, without being any less capital intensive, therefore it’s reasonable to infer that semiconductor production accounts for a significantly higher proportion of the NAICS 334 growth. Following from that, another reasonable inference that most of the Computer/Electronic/Electrical construction spending was likely concentrated in the NAICS 334 Computer and Electronic product segment, relative to NAICS 335.

The data suggests that investment in semiconductor manufacturing structures outpaced both IRA-sensitive categories and the broader ecosystem. Based on total capital expenditures reported in the 2021 and 2022 ACES, structures capex for computer and electronic product manufacturing increased by 39%, adding up to a $3.3B gain. Structures capex within the more IRA-sensitive NAICS 335 manufacturing subsector grew by an impressive 72% from 2021 to 2022, but because growth in NAICS 335 structures capital expenditures materialized from a lower base, it only translated to a $528M gain, roughly a sixth of the additional structures capital spending in the CHIPS-sensitive NAICS 334 manufacturing subsector. These categories outperform the entire manufacturing sector, which saw 11.5% growth in structures capital spending.

Analysis of spending by census division shows policy effects

From the twelve months ending December 2019 to the twelve months ending October 2025, total manufacturing construction spending rose from $80.5 billion to $224.2 billion — an increase of $143.6 billion.6 The regional concentration of the manufacturing construction increase is even more striking.

The Mountain census division — encompassing Arizona, where TSMC and Intel (among the largest CHIPS award recipients) have committed to major fabrication facilities, as well as Idaho (Micron), Utah (TI), and Colorado (Microchip, Entegris) — accounts for $41.5 billion of this increase, or 28.9% of the national gain. For a census division that represented just 6.5% of national manufacturing construction, the past six years have been transformational. The division now represents nearly 21% of all construction spending on manufacturing facilities, and represents nearly 30% of the gain in manufacturing construction since 2019. The Mountain division is not a legacy manufacturing region, and its outsized contribution to the construction boom is strongly attributable to semiconductor megaprojects that were explicitly positioned to benefit from CHIPS Act awards.

Three other regions saw noteworthy manufacturing construction spending booms over this same period:

• The South Atlantic division accounts for $37.8 billion (26%), but the region also benefited from a battery- and EV-related investment alongside CHIPS-supported projects in North Carolina, Virginia, and Georgia.

• The West South Central division — predominantly Texas, home to Samsung, Texas Instruments, GlobalWafers, and other CHIPS recipients — contributed $22.1 billion (15.4%).

• The East North Central division, where Intel’s $28 billion Ohio campus and SK Hynix’s Indiana packaging facility are located, added $24.6 billion (17.2%).

These four census divisions account for 88% of the national increase in manufacturing construction spending since 2019. The remaining five divisions — including the heavily industrialized Middle Atlantic and New England regions — contributed almost nothing to the aggregate gain, and the Northeast, as a whole, actually shows a decline. Taking into account planned CHIPS-funded developments in New York, we should soon see some uplift in the Middle Atlantic’s construction spending numbers.

Manufacturing construction spending by census division (trailing 12-month sums, billions). The Mountain, South Atlantic, West South Central, and East North Central divisions dominate the post-2021 increase.

Alignment between CHIPS awards and census data

Comparing CHIPS awards against actual Census Bureau construction spending gains by census division provides a useful, if imperfect, alignment check. The dollar value of CHIPS awards skews to a narrow set of census divisions, likely due to major megaprojects, and those divisions also constitute a disproportionate share of growth, suggesting policy impact on construction spending.7

The Mountain and West South Central divisions show the highest alignment between CHIPS awards and the broader manufacturing construction boom and the predominant boom in the Computer/Electronic/Electrical segment specifically. These two regions represent 44% of the growth in manufacturing construction spending since Calendar Year 2019 and were outsized recipients of CHIPS awards. TSMC, Intel, and Micron made major investments across Arizona, New Mexico, and Idaho, while Samsung made historic investments in Texas. While there may have been other investments of relevance in Texas, including Exxon and Chevron’s petrochemical facilities and Tesla’s gigafactory in Austin, the semiconductor manufacturing investments were likely a major cause of that spending growth in the Mountain census division.

As noted above, not all of the boom in manufacturing construction spending, or even in the manufacturing of Computer/Electronic/Electrical products, should be attributed to CHIPS. There are many regions, including the South Atlantic and East North Central, that saw a surge in manufacturing investment, but where CHIPS was unlikely to be the sole or even predominant driver.

The East North Central division was a substantial recipient of CHIPS Awards, most notably for the Intel facilities planned in Ohio. These facilities have faced ongoing delays, so while the region ranks highly in both manufacturing construction spending and relevant CHIPS awards, the policy effect is still relatively murky compared to other census divisions. This could possibly reflect the fact that CHIPS awards are not fully disbursed until the achievement of certain milestones.

The Pacific, West North Central, and East South Central divisions have seen some growth without any comparably significant CHIPS awards, but with other identifiable catalysts such as the advanced manufacturing investment tax credit, IRA-linked projects in the battery sector, and efforts to onshore pharmaceutical manufacturing. The South Atlantic division experienced a significant boom, representing 26% of growth over the past five years, the second highest of all the census divisions. That boom is tied to battery manufacturing that triggered construction spending very soon after IRA and CHIPS were enacted. In total, the regions with limited CHIPS awards make up 41% of the manufacturing construction spending boom.

An absence of CHIPS investment in a division corresponded to less growth in manufacturing facilities, providing marginal evidence in favor of a CHIPS policy effect. The Northeastern divisions (Middle Atlantic, New England) have not seen CHIPS awards for projects that commenced prior to 2025, and have seen no growth in manufacturing construction spending through this boom.

These geographic trends are likely to evolve over the coming years, and especially so if CHIPS has a pronounced effect. There are awards tied to Micron’s investments in New York where construction has only just begun. The performance of manufacturing construction spending in the Middle Atlantic division in the coming years will provide another opportunity to identify and analyze the effects of CHIPS investment, albeit against a different set of confounding variables such as tariffs or changes to tax treatment such as bonus depreciation.

Quantification of CHIPS effects

The skewed regional variation in CHIPS Awards and manufacturing construction spending, along with the predominance of the Computer/Electronic/Electrical segment, gives us a guide for benchmarking the potential size of CHIPS effects.

Relative to a static baseline of 2019 manufacturing construction spending, cumulative spending has outperformed by $433B as of the latest data through October 2025. In other words, relative to merely replicating the same level of spending and implied activity of 2019, the total additional dollars of spending devoted specifically to the construction of manufacturing facilities was $433B over the past 70 months. This approach accounts for some of the local cyclicality in construction spending without over-indexing on a local high watermark in construction spending, especially if only fleeting. While this cumulative estimate misses the role of rising inflation in construction costs, so long as the relative construction costs across manufacturing subsectors and US regions were roughly comparable, they should not materially affect the analysis.

The Computer/Electronic/Electrical segment makes up the vast majority of this $433B cumulative outperformance at roughly 76% (~$329B). The two census divisions where the nexus between CHIPS awards and construction spending are most apparent — Mountain and West South Central — represent 47% of the $433B cumulative increase in manufacturing construction spending, approximately $203B.8 Keep in mind that the top three divisions saw meaningful CHIPS awards but also saw manufacturing investments for other reasons. Meanwhile, other divisions with less significant CHIPS awards still saw meaningful upticks in manufacturing investment.

If we take the 47% spending share of the top two regions as a benchmark for a charitable estimate of plausible CHIPS-attributable spending, and then apply the national Computer/Electronic/Electrical share of 76% to account for the fact that not all construction in those regions was in CHIPS-relevant industries, we arrive at 47% × 76% ≈ 35%, or roughly $152B of the $433B total manufacturing construction spend. This gives us a soft negative inference: the direct policy effect from CHIPS alone was probably no greater than 35% of the cumulative outperformance in manufacturing construction spending. Other policies, including those under the IRA, likely contributed to the broader boom, and other market forces and state-level competitiveness considerations surely played some role in the scale and allocation of investment.

If CHIPS accounted for even a tenth of the increase in manufacturing construction spending over the past five years (~$43B), that would still exceed what Congress appropriated and what the CHIPS For America Fund allocated in financial assistance. As more data is released, we will better understand how awards translate into construction spending. More importantly, the construction activity tied to these projects is only the first layer of relevant economic output and return on these investments in semiconductor manufacturing. The full cost-benefit analysis of CHIPS must go well beyond an analysis of construction spending, but the early returns already suggest meaningful economic effects that helped to crowd in private capital expenditures in structures.

Considering the counterfactual

This analysis is not motivated by the question of whether the CHIPS Act was the sole cause of the manufacturing construction boom — it plainly was not. Post-COVID demand shifts, the Inflation Reduction Act, supply chain security concerns, and broader reshoring pressures may all have helped spur manufacturing investment. The question is how much of the observed surge would have materialized absent specific policy intervention. While none of the individual pieces of evidence is dispositive on its own, taken together they suggest a high likelihood of positive and economically meaningful policy effects.

The evidence examined here suggests that the counterfactual boom would have differed from what occurred in at least three respects. In scale, it would likely have been more modest: firms may have pursued incremental capacity expansions and brownfield investments rather than the greenfield megaprojects that dominate the construction spending data. In location, it is unlikely to have concentrated so heavily in the Mountain and West South Central divisions where CHIPS awards and megaprojects were so prominent. In composition, it would have been less dominated by the Computer/Electronic/Electrical segment.

The FEDS Note’s observation that planning surged before enactment is factually correct, but leaves open the question of whether the surge was in response to policy developments or in anticipation of them. While the former may be intuitive for skeptics of the CHIPS Act, the evidence is more consistent with firms treating construction planning as an option to be well-positioned in the event of policy enactment. For the timeline shown by the FEDS Note to invalidate strong policy effects, CHIPS’s influence must have only taken hold in August 2022 — an assumption that ignores the legislative history, sectoral skew, and geographic distribution of spending and CHIPS awards.

The more complete story is that policy expectations likely began shaping firm behavior well before enactment. Firms were willing to commit to planning expenses and even break ground for the option to scale up if legislation passed. The cumulative effect suggests that anticipated and enacted legislation catalyzed investment at historic scale and in locations that market forces alone cannot explain. The resulting construction boom bears the fingerprints of key industrial policies more clearly than a surface reading of the timeline might suggest.

I found that my perspective had shifted when I picked the book up again a few weeks ago. Wilson’s analysis suddenly seemed like an essential lens for understanding the last five years of US industrial policy under both the Biden and Trump administrations.

This piece deploys both Wilson’s account and my own experience at CHIPS to offer four perspectives on the past, present, and future of US industrial policy:

1. When national security demands rapid industrial capacity expansion, supply-side public investment appears essential — a pattern evident in WWII, reinforced in CHIPS, and visible across successful industrial buildouts from postwar Japan to contemporary China.

2. The American tradition of public-private partnership has worked through a distinctive dynamic: keeping government and industry as separate spheres while building genuine collaboration and maintaining productive (if at times fierce) commercial tension.

3. Recovering the state capacity that fueled the WWII mobilization — both by removing procedural barriers and creating a political culture that values public service — is essential for advancing national security today.

4. Today’s geopolitical context demands fundamentally different approaches to executing industrial policy — particularly when it comes to demand-side interventions.

It’s become something of a trope in the current policy discourse to say that competing with China requires becoming more like China. But this obscures the fact that we have an American tradition of industrial policy — one that can illuminate our current moment but remains underappreciated. In reviewing the history, I consider how and why we moved away from this tradition after WWII. Understanding the reasons for that departure may be just as important as understanding the tradition itself.1

Subscribe now

Supply-side investments are essential

When national security demands rapid industrial capacity expansion, supply-side public investment appears to be essential. This was true in World War II. It was true in CHIPS. And examining other major industrial buildouts over the last century, the pattern holds. Demand-side policy alone, even with the clearest possible market signal, has not historically been deployed to drive strategic industrial buildouts at speed and scale. This is not proof that demand-side policy alone could never work. But the absence of successful examples should counsel caution about relying on it exclusively.

The scale of government supply-side investment during World War II was staggering. The industrial boom across munitions, explosives, ships, tanks, aircraft, and upstream inputs was driven by so-called “GOCO” facilities — government-owned, contractor-operated. Under these arrangements, the government financed and owned the entire investment, contracting out construction and operations to a manufacturer, who earned a fee. The government spent close to $20 billion on manufacturing facilities and machinery over the course of the war, equivalent to more than $2.5 trillion today as a share of GDP. This was more than double the amount of private sector investment. Public capital had financed about one-tenth of the value of American war plants during World War I; as GOCO facilities proliferated during World War II, it financed two-thirds.2

By 1945, the government owned nearly all national production capacity in aircraft, guns and ammunition, shell and bomb loading, ships, synthetic rubber, and enriched uranium and plutonium. Nearly half of this investment came directly from the War and Navy Departments, with the other half flowing through the Defense Plant Corporation (DPC) — a subsidiary of the Reconstruction Finance Corporation, which had served as a source of New Deal financing since 1932. By war’s end, the federal government owned close to a quarter of the value of all the nation’s factories — the largest collection of state-owned assets outside the Soviet Union.

Similarly, the CHIPS program was a supply-side intervention. The program consisted of a 25% investment tax credit alongside $39 billion in grants, which typically amounted to an additional 10% of investment. Some, including President Trump and Secretary Lutnick, have implied that supply-side incentives are wasteful, and that demand-side tools like tariffs alone would encourage buildout. But this argument was available to our World War II predecessors in even starker form. They didn’t need tariffs to create demand; the military’s procurement signal was massive and unambiguous. Yet the mobilization’s success relied on massive government supply-side support nonetheless.

Despite the skepticism, both the Biden and Trump administrations have acted in accordance with this historical lesson and favored supply-side interventions. The Biden administration instituted the 25% tax credit and roughly 10% in additional grant funding per CHIPS project. The second Trump administration has actually expanded the tax credit to 35%. Whatever the rhetoric, the underlying policy has seen bipartisan continuity: when national security demands rapid industrial capacity expansion, supply-side public investment has been deemed essential.

This is not to diminish the importance of demand-side policy. Government procurement drove the WWII mobilization, and the lack of robust demand-side tools presented real challenges for our efforts to implement CHIPS. As discussed below, today’s context requires more creative policy approaches on the demand side, alongside the continued need for supply-side investment.

The central role of supply-side investment extends beyond the American experience. Japan’s Ministry of International Trade and Industry directed strategic industrial development with substantial state financing. South Korea channeled subsidized credit to advance its industrial aims. Taiwan built designated industrial parks and deployed state capital (alongside private investment) to launch TSMC. Singapore’s Economic Development Board coordinated industrial strategy and sponsored manufacturing investments. China has sustained massive state investment in infrastructure and strategic industries. The mechanisms and magnitudes of supply-side investments have differed, but each case reflects a conviction that public investment could shape industrial outcomes that markets alone would not produce.

The historical record — American industrial mobilization in WWII, the East Asian development successes, China’s rise — offers no clear examples of strategic buildouts in capital-intensive manufacturing that didn’t rest on state investment. Public support has been foundational, not peripheral. This is not to claim that supply-side investment alone caused these outcomes — the counterfactual is unavailable. But its presence in every successful case is difficult to dismiss as coincidence. The lesson from Wilson’s account, reinforced by CHIPS and international experience, is that supply-side public investments have been at the core of the most successful industrial expansions in recent history.

Public-private partnerships have succeeded through separation and tension

How did the partnership between government and industry actually work during World War II? The relationship was a complex web of interlocking and overlapping functions. Government provided the bulk of supply-side capital, but industry made critical investments too. Industry built and operated the vast majority of GOCO facilities, but the government directly ran operations in some sectors (most notably, shipbuilding yards were operated entirely by federal agencies). Government demand was the dominant force across the economy, but commercial demand continued to shape sectors like oil and steel, where civilian and military needs overlapped. Industry generally managed its own supply chains, but government actively steered equipment and materials across the industrial base.

Yet for all this overlap and entanglement, public and private entities remained fundamentally distinct. It was not like modern-day China, where CCP committees operate inside major companies — even nominally private ones like Huawei. Rather than fusing into one, lines were maintained between the American government and industry, with relationships defined by contract. The government acquired property or other assets and contracted with businesses for operation. It also provided businesses with loans and insurance, which then carried out production efforts independently. Regardless of the contractual relationship, government and industry remained distinct spheres with different institutional interests, objectives, and constraints. This separation would prove fundamental to the function of the partnership.

Building the partnership

Despite the institutional separation, government and industry developed effective working relationships. Roosevelt deliberately cultivated ties, placing industry titans in central roles: developer Jesse Jones to run the Reconstruction Finance Corporation, GM’s William Knudsen to coordinate production, and Sears executive Donald Nelson to lead the War Production Board. He also appointed Republicans to the military’s top civilian posts — Frank Knox, a former GOP vice-presidential candidate, as Secretary of the Navy, and Henry Stimson, a Wall Street lawyer and former Secretary of State under Hoover, to run the War Department. As Wilson notes, Roosevelt was willing to let “enemies of the New Deal run some of the most powerful offices in the wartime federal government.”

Other historical accounts of the period, like Chris Hughes’s Marketcrafters and Arthur Herman’s Freedom’s Forge, chronicle how figures like Jones and Knudsen relentlessly worked their industry connections to advance the mobilization effort. These weren’t just symbolic appointments — they provided the relationships and trust essential to making the public-private partnership work at the operational level. Even New Deal stalwarts like Harold Ickes (who led coordination with the oil industry) developed strong working relationships with their industry counterparts.

Commercial tension

Yet for all of the genuine collaboration and mutual respect, the relationship between government and industry was marked by constant tension, often boiling over into outright hostility. This wasn’t an unfortunate side effect. It was structural and practically inherent to the American model. Government and industry remained distinct entities with different institutional interests, operating in constant commercial relationship — that separation generated real friction.

Business leaders constantly criticized the government, and publicly. Oil executive J. Howard Pew, who once gave Ickes a standing ovation at an industry dinner, told business leaders in 1943 that the wartime mobilization had succeeded despite “the schemes of bureaucrats and economic planners.” George Romney, who led an industry association for car companies and later became governor of Michigan, called Washington’s “paper blizzard” reminiscent of “Hitler’s methods.” NAM president Ernest T. Weir railed against “whimsical restrictions of bureaucrats.”

As Wilson tells it, some of this tension reflected frustration with government dysfunction — paperwork mountains and bureaucratic overreach that made the public servant in me cringe. But much of the tension was driven by government competence, not incompetence. The state developed deep “economic knowledge” and “regulatory flexibility and capability,” and used them aggressively. Profit margins were monitored and controlled through pricing. Statutory excess profits taxes reached 90%. Contract terms were renegotiated when companies’ costs came in below projections. As Wilson observes, “Far from being a blind behemoth or price-insensitive pushover, the wartime state often acted as one tough customer.”

Wilson’s deeper insight — and I’m glossing here, as this is my interpretation of his account — is that this tension was productive. It kept both sides honest. The government prevented profiteering and kept costs down. Industry pushed back on government inefficiency and overreach. The collaboration worked not despite the friction, but because of it.

Lessons for today

Reading this history, I sometimes felt like the CHIPS team was an unwitting heir to this American tradition. Our role was fundamentally different from the government’s role in WWII — we weren’t functioning as customers, and our leverage was more limited. We were providing incentives covering perhaps 10% of project capital (alongside the 25% tax credit) rather than owning the facilities outright. But certain dynamics felt familiar.

We maintained the separation between government and industry. We developed deep relationships with industry, buttressed by the commercial credibility of our team members — many of whom had industry and finance backgrounds.

But there was also constant tension. Applicants pushed back when we overreached. That friction improved us — it forced discipline that made the program better. But most of the tension wasn’t about bureaucratic irritation. It was about dollars, milestones, and other protections for taxpayers. We were pushing companies to do more for less than they wanted to.

Early in the program, a former senior policymaker told me: “If Pat Gelsinger is totally happy with how things go, you probably haven’t done your job.” That stuck with me. When public and private remain distinct, when they’re commercial counterparts rather than fused entities, tension is baked into the model. And based on what Wilson documents, that’s a feature, not a bug.

But I worry that today’s environment threatens aspects of this model. While I think there are real use cases for government equity, the current administration’s haphazard approach — while not amounting to state-industry fusion (the Intel stake, for example, carries no control rights) — could blur the public-private distinction over time. More concerning to me is the risk that productive tension devolves into corporate acquiescence. If companies perceive that disagreeing with the government might affect their standing with the government, the incentive structure changes (and recent disputes with firms like Anthropic suggest this isn’t merely hypothetical). Wilson documents that WWII business leaders could publicly attack government policy without consequence. That freedom to criticize, grounded in institutional separation, was central to the American tradition. When it erodes, we’re departing from norms that have historically distinguished our approach.

Building state capacity at speed

When national security demanded it, the WWII-era government built state capacity with astonishing speed — both in expanding the bureaucracy itself and in producing results. Recovering that capability today requires confronting two challenges: the procedural constraints that have accumulated over decades, and a political culture that no longer consistently values public service. Without addressing both, we risk lacking the state capacity that national security may require.

The wartime government built massive bureaucracies at an extraordinary pace. The War Production Board employed 25,000 people. The War Department had 150,000 procurement officers spread across multiple agencies. Treasury grew from 45,000 to 95,000 employees as the federal government applied a broad-based income tax for the first time. By comparison, the 180-person team we built for CHIPS seems downright quaint.

But more impressive than bureaucratic scale was the speed at which the state produced results. In June 1940 — nearly 18 months before Pearl Harbor, before the crisis of war was fully at hand — chemical company DuPont agreed to build and run a smokeless powder facility on a 5,000-acre site in Indiana, WWII’s first major GOCO facility. Production (not construction — production) began less than one year later in spring 1941, while 27,000 people were still building out parts of the plant. The facility was fully scaled by the second half of 1942, employing 9,400 workers and producing a million pounds of powder per day. By the time of Pearl Harbor in December 1941, 17 explosives and munitions plants were already up and running with 32 more in the pipeline. By the end of the war, there were 73 GOCO plants in this sector alone employing 400,000 people. All of this happened at a pace that seems almost incomprehensible today.

Perhaps just as telling as the stories themselves is what’s absent from them. Procedural constraints on state capacity and barriers to construction simply don’t appear in Wilson’s narrative. When I recall our struggles to navigate procurement rules, federal hiring processes, NEPA reviews, contractual protections for taxpayers, the Paperwork Reduction Act, the Davis-Bacon Act — or when I think about the layers of permitting that TSMC had to work through to get its first Arizona fab off the ground — it doesn’t feel like we’re talking about degrees of difference in efficiency. We’re looking at fundamentally different systems. At CHIPS, we mostly powered through these procedural constraints to deliver on our program’s objectives — but, as we’ve explained before, the effort required didn’t feel sustainable or replicable.

To the extent that Wilson’s account has an ideological thrust — and it’s subtle — it’s this: national security requires a state energized and entrusted to solve big challenges. For me, that necessitates taking a hard look at the procedural constraints holding government back today. I think the emerging discourse around state capacity has a national security dimension that needs amplifying. When semiconductor supply chains or defense industrial base challenges demand speed, procedural barriers become strategic vulnerabilities.

But procedural reform, while critical, isn’t sufficient on its own. Wilson’s narratives show hundreds of thousands of civilian and military public servants working tirelessly as part of a celebrated national effort. This points to something equally essential: a political culture that values public service.

The current administration’s initial approach to federal workforce reduction illustrates the danger. Increasing the performance and efficiency of government is undoubtedly a worthwhile objective. But when workforce reductions are executed indiscriminately and paired with rhetoric that denigrates federal workers broadly, it creates risks for national security capacity. The immediate consequence has been talent loss — the CHIPS program alone lost dozens of capable public servants. The longer-term impact on recruiting may be more severe. If potential public servants no longer view federal employment as stable or the work as valued, they will be less likely to serve. This is particularly concerning for programs that require both technical expertise and commitment to public mission.

Wilson’s account suggests we once knew how to build state capacity quickly when national security demanded it. Whether we can recover that capability depends on both removing procedural barriers and rebuilding respect for the work of governing itself. National security requires attending to both.

Subscribe now

This time is different

So far I’ve focused on the ways that Wilson’s history offers lessons that feel directly relevant to the industrial policy challenges we face today. But the WWII history can only take us so far, because the world we’re navigating is fundamentally different.

A different strategic context

Consider one final story from the war: When Japan seized Southeast Asia after Pearl Harbor, it created a supply chain crisis that could have lost us the war — capturing about 95 percent of the world’s natural rubber supply. The response was dramatic: an all-out mobilization to build a synthetic rubber industry from scratch. The DPC spent $700 million building three dozen facilities. By war’s end, synthetic rubber production reached three hundred times the level in 1940.

Two things stand out about this episode. First, how existential the threat was. Rubber was life or death for the war effort. When the government sold off surplus war property after 1945, they held onto the rubber plants longer than almost anything else, keeping most of them in strategic reserve until the mid-1950s. Second, the rubber appears to have been unique — the only major supply chain vulnerability to a hostile power in the war.

Today’s situation couldn’t be more different. The American economy is more than 10 times larger in real terms than it was in 1945. But more importantly, the global economy has become vastly more integrated and complex. Our chief geopolitical competitor manufactures one-third of global output today — likely approaching 40% by decade’s end. Supply chain dependencies have proliferated far beyond any single commodity. Semiconductors and rare earths come to mind as modern parallels to the rubber challenge, but vulnerabilities extend far beyond these obvious cases. Pharmaceuticals present another clear dependency. And given the complexity and opacity of modern supply chains, we genuinely don’t know how many such challenges we have.

This different strategic context demands a different conceptual framework for industrial policy. Industrial policy today isn’t just about making weapons for war — though that capacity remains critical. It’s part of a broader geopolitical competition in which we’re trying to establish leverage and mitigate our adversaries’ leverage across a wide range of economic activity.

Three key differences in execution

The shift in strategic context creates significant differences in how industrial policy must be executed today compared to WWII. Reading Wilson’s account, three such differences stood out to me.

First, as Kurt Campbell and Rush Doshi have articulated, we cannot build resilient supply chains to compete with China without relying on other countries. WWII mobilization was overwhelmingly a domestic effort — American government working with American companies to build American capacity. Today, coordination with allies is essential. Building resilience requires coordinating industrial strategy across allied economies in ways that didn’t arise during WWII. This adds layers of diplomatic and economic complexity that have no clear historical precedent. Moreover, recent ruptures in our traditional alliance frameworks complicate this imperative further. We haven’t fully reckoned with how to achieve global industrial coordination in light of the shifting global order.

Second, both capital flows and the firms that direct them are now global. In WWII, the government worked almost entirely with American companies to drive American investment. Today, we are competing for incremental global capital investment. We need to attract not just American companies but foreign ones — and we’re competing against other countries doing the same. At CHIPS, attracting foreign investment was central to our strategy — we couldn’t afford to be narrowly nationalist about onshoring. The goal was bringing semiconductor manufacturing capacity to the United States, regardless of whether the company was American or foreign.

Third, the role of government demand has fundamentally changed. In WWII, government procurement was everything — the military was the customer for virtually all industrial output. Today, government demand plays too small a role in many critical industries to serve as the primary driver. The semiconductor industry alone is rapidly approaching annual revenues of $1 trillion globally. Government procurement, while important in specific segments, cannot drive investment at that scale the way wartime military demand did in the 1940s.

Getting creative on demand-side tools

In the absence of procurement as the primary lever, policymakers have been developing creative approaches. The Trump administration’s recent rare earths deals illustrate one direction: using the government balance sheet to create demand-side incentives without direct procurement. Offtake guarantees and price floors can reduce investment risk and accelerate capacity buildout.

We considered similar approaches for CHIPS but found they mostly didn’t make sense in the semiconductor context. The scale is simply too large to be an efficient use of government resources, and the products are too differentiated — hundreds of different chip types with distinct specifications and customers. Moreover, the relationship between manufacturer and customer is too integrated and technically complex to allow meaningful government intermediation.

Tariffs and domestic sourcing requirements are other obvious demand-side instruments. But given how integrated supply chains have become, both need to be wielded carefully. Applied poorly, they can easily disrupt existing production and do more harm than good.

One underexplored approach, in my view, is direct regulation of industrial inputs. Sometimes the simplest answer is the most effective: if we don’t want industry to depend on our adversaries for certain inputs from China, the government can seek to regulate that dependence directly. We’ve taken some steps in this direction — for example, the federal government just proposed a rule to implement a provision of the 2023 NDAA that requires federal vendors to ensure their products contain no chips from designated Chinese firms — but in my view, this is an area that deserves much more exploration.

Charting the future by rediscovering the past

Perhaps the strongest lesson from Wilson’s account isn’t about any specific industrial policy tool. It’s that active government intervention in industrial capacity isn’t foreign to the American tradition — it is the American tradition, particularly when national security has demanded it.

In the decades following World War II, the country moved away from that tradition. A political consensus emerged that treated industrial interventions as inconsistent with the American way rather than part of it. There are a few reasons for that shift.

One is revealed in painstaking detail in Wilson’s account: an energized and organized effort by industry itself to extol its own role and diminish the government’s role coming out of the war, as part of a broader campaign to shift American politics away from the New Deal. Wilson argues that these efforts to diminish the role of the state won the narrative and shaped politics going forward, diminishing the state’s role in industry for decades to come.

But I also think two broader ideological shifts were at play. First was an economic consensus that emerged in the 1980s embracing trade liberalization and minimal state intervention in the economy. Second was a foreign policy consensus that economic integration and interdependence would promote peace and security — that deeper trade and investment ties between nations would make conflict less likely and advance democracy globally. Together, these created a powerful logic: global economic integration would generate prosperity through free markets while simultaneously reducing geopolitical risk.

Today, there’s active debate on the economic case for global integration — whether the efficiency and growth it generated justified the dislocations it created. But regardless of the economic debate, there’s a growing consensus as a matter of national security that economic interdependence has created vulnerabilities that need to be addressed through industrial policy.

That consensus is a response to the world as it is today — particularly the reality of China’s rise as a manufacturing and technological superpower. But recovering industrial policy for national security isn’t a break from American tradition — it’s a return to it. And that history should fuel confidence: we’ve done this before, and we’ve done it well.

The CHIPS statute did not mandate a detailed milestone regime, which allowed us to devise our own approach. CHIPS milestones were not the check-the-box, activity-oriented progress payments typical of government grants. In practice, milestones became our most effective tool for enforcing accountability and shaping outcomes. We defined customized, enforceable commitments tied to the results we actually cared about — not just fabs built, but output, technology readiness, customer acceptance, financial viability, national security commitments, and supply-chain resilience.

For us, milestones were policy instruments for shaping industrial outcomes rather than mere safeguards for appropriations.

Building an outcomes-oriented mindset

Most traditional government grant programs disburse funding based on standardized cost- or activity-based “milestones” that prioritize compliance and process over outcomes. For example, they may be procedural (like those tied to the OMB Uniform Guidance) and related to demonstrating that allowable costs have been incurred, or that certain activities have been fulfilled, such as entering subrecipient agreements or executing a contract. This approach can validate progress and guard against fraud or mismanagement, but it does not help steer program outcomes.

To be fair, there are a few major programs that engage serious private sector companies and use milestones to drive accountability. Examples include DOE’s Loan Program Office, ARPA-E, and NASA’s Space Act Agreements. In these cases, the government funds some combination of execution, scale-up, technology, or visible failure risk; the milestones aim to mitigate those risks and ensure operational readiness.

CHIPS took this to a new level. Beyond mere accountability and risk mitigation, we used milestones to shape strategic behavior and build ecosystem resiliency. We wanted to both manage project risk and drive national strategy alignment.

There were other tools available to us for taxpayer protection, like the completion clawback mandated by the statute:

“...the Secretary shall 1) determine target dates by which a project shall commence and complete, and 2) set these dates by the time of award. If the project does not commence and complete by the set target dates … the Secretary shall progressively recover up to the full amount of an award.“1

Most grant programs would have relied on this clawback, paired it with standardized progress payments (such as a certain percentage of allowable capital expenditure spent), and called it a day. Instead we chose the harder — and more consequential — path. We viewed the statutory clawback as the hook that allowed us to build a much more ambitious milestone framework. But designing customized results-oriented milestones requires significant upfront effort and industry expertise, and puts more weight on post-award monitoring.

Rather than treating milestones as an afterthought, we began deliberating on them early in the award process. Before any project reached our Investment Committee (IC), it went through multiple rounds of review in our twice-weekly, multi-hour CIO/CSO (Chief Investment Officer/Chief Strategy Officer) meetings. The core investment team would present to me, my Chief of Staff, our Chief Economist, our Chief Strategy Officer, and a panel of senior semiconductor and national security experts from our strategy team. As we approached recommending an investment, the team had to present a page on milestones, which was subject to a set of core questions:

• Which milestones will be critical to this project and why?

• What are the risks we are concerned about and how can we use milestones to protect against them?

• What would real success look like for this project or company and how can we use milestones to incentivize that outcome?

• What could this company uniquely contribute to the overall ecosystem and how could milestones draw that out?

These were always my favorite meetings, because we rigorously debated and challenged one another on these questions. Revisiting them week after week, over months, built a culture focused on shaping outcomes and set high expectations for the team. We didn’t simply accept what companies told us — we vetted it.

These deliberations slowed down negotiations and announcements. In our preliminary memorandum of terms (PMTs), we decided to negotiate the details of the milestones, while using boilerplate language for most other commitments (e.g., workforce, environmental, cybersecurity), postponing those decisions to final documentation.2 Our milestone discussions were often the most challenging conversations with applicants — on par with negotiations over total funding. But because they were critical to driving outcomes, we wanted to reach agreement before making any announcements.

Subscribe now

Our layered milestone strategy

The CHIPS milestones could be categorized into three buckets and goals:

1. Protect against the obvious downside risks. These milestones focused on construction, technology, and production: Is the facility built? Is it operational? Does the technology work? These were table stakes and classic taxpayer protection.

2. Incentivize desired commercial and financial outcomes. The next layer of milestones focused on commercial and financial viability. We wanted to avoid scenarios where a company built operational facilities, but lacked customers, commercially sustainable yields, or enough capital to sustain the facility. These were company-specific risks that needed a customized approach. The milestones in this category addressed downside risk, but with a more offensive, results-oriented objective.

3. Drive outcomes related to national security, supply chain resiliency, or overall ecosystem health. These milestones were pure offense, and the most differentiated aspect of our approach. Beyond the individual project, we considered what a particular company could do to further overall programmatic goals, such as bringing more of the supply chain to the US, or committing to other technologies or products critical to national defense.

This combination of milestones required specificity and robust post-award data collection and monitoring. We needed to build deep semiconductor industry expertise and company-specific knowledge to be able to design them and verify that they were achieved. And it required us to establish deep, ongoing relationships with recipients to surface problems early rather than after commitments were missed.

Bucket 1: Protect against the obvious downside risks

The fundamental goal of milestones was to get facilities built and operational as quickly as possible. While this seems straightforward, “built and operational” is open to interpretation. For example, Intel had a widely known “shell-first strategy” in which they would build the fab structure, including installing and sealing the cleanroom (the critical space in a fab where sensitive manufacturing can occur without risk of contamination), but then pause before installing tools (which typically amount to 70-75% of the total cost of a leading-edge fab) or producing anything. Would this satisfy our definition of “built”? Would we have been considered successful if we stood up multiple shells across the country, but without tools or production? For us, the answer was no. In most cases, success required installed and operational tools, with wafers and chips coming off the line.

Our work therefore required building a detailed understanding of each company’s operational plan and translating that into customized, verifiable stage-gates for disbursement. This approach called for extensive early engagement with applicants, and deep industry expertise. It also kicked up considerable internal debate about timelines, grace periods, and how we would handle waivers if milestones were not met (whether due to construction delays or lack of demand).

CHIPS milestones were tied to specific events and conditions: shell completion, cleanroom sealing, tool installation, defined levels of wafer production capacity, and production volume.3 We also tied milestones to proof of technology capability, such as proof of tech-transfer from an overseas fab. This required the company to prove the transfer of process databases, tool configurations, and other components necessary for manufacture, and demonstrate that the chips could now be produced in the US.

One additional design choice proved especially important in negotiations. We added language to the NOFO that signaled the customized nature of these milestones and stated that “The rate of disbursement is generally expected to be proportional to the rate at which non-Federal dollars are expended over the course of the project.” This set the expectation that the government would fund a consistent percentage of total capital spent at each milestone. Establishing this expectation created an important negotiating tool for us. While we had little flexibility on total award size, we could pull forward disbursements; by placing a higher percentage of funding on earlier milestones, we increased the award’s effective value and shifted negotiations from the headline amount. The trick was to find the right balance between earlier funding and driving accountability in later milestones.4

Bucket 2: Incentivize desired commercial and financial outcomes

Even with facilities built, tools installed, and production underway, a project could still fail commercially or financially.

To manage that risk, our second layer of milestones focused on viability. We asked ourselves what success would look like for a given project, and used our milestones to encourage that outcome. We had to game out the following scenarios:

• If the facility installs equipment that has the potential to produce at a stated capacity, can the company actually produce at required technical standards and economically competitive yields?

• If a facility is completed, is there customer demand for its output? Are customers qualifying products? At what stage? Are there committed volume contracts?

• Is the company financially viable? Do they have the required capital to achieve profitability at the fab over time?

These milestones also required customization. If we were concerned about the viability of the technology, one option would be to align milestones with the targeted yields the companies share with their customers. Or we may have required documented volume commitments from customers for a certain amount of output. For smaller or less well-capitalized companies, we created explicit financial milestones, such as raising outside capital or refinancing debt, which had to be met to release additional funding. Companies could then use our funding commitment to engage with customers and/or financial partners to catalyze those outcomes. This approach allowed us to support companies earlier in their lifecycle while still protecting taxpayer capital and forcing proof points before scaling.

There are a few milestone examples that have been well publicized. Intel reported in its 2024 10-K that their agreement “contains detailed milestones we must achieve for us to receive the funds, including the achievement of various milestones with respect to capital expenditures, facility completion, process technology development, wafer production, Intel products insourcing, and external foundry customer acquisitions.” Unlike most leading-edge semiconductor companies, Intel is vertically integrated and really consists of two businesses. Its products business designs the chips inside most PCs and is similar to NVIDIA and AMD; its manufacturing business, which is more akin to TSMC, is responsible for producing chips for Intel products (and, in the future, for other companies). Only the manufacturing business is important for our national security interests, so our milestones focused on that business demonstrating success with both Intel’s own products and external customers. Unfortunately, when the Trump administration converted the Intel award to an equity stake, our award agreement was nullified and the milestones holding Intel accountable disappeared (and its manufacturing business lost over $10 billion in 2025).

Another notable example was our negotiation with Wolfspeed, a US silicon carbide company creating power semiconductors for electric vehicles, industrial power, and data centers. We announced a PMT with them in October 2024, but did not reach a final agreement before the end of the Biden administration due to their challenging financial situation. The company’s announcement of their PMT details the financial milestones we set when financial viability was in question:

“...funds will be conditioned upon the achievement of certain operational and construction milestones and other requirements. The PMT includes an obligation for Wolfspeed to raise an aggregate of $750 million in debt financing … [It] also requires Wolfspeed to undertake further actions with respect to its capital structure, including (a) restructuring or refinancing its … convertible notes; (b) deferring … cash interest payments … and (c) raising up to $300 million in additional capital.”

In this case, we structured milestones to ensure no federal dollars would be disbursed unless Wolfspeed resolved its financial situation. We could have foregone the engagement altogether given the circumstances, but their technology was critical for national and economic security. We wanted to use our milestones to encourage a successful outcome by signalling to the market that, if the company could meet these milestones, there was a strong possibility of an award. Our announcement could potentially catalyze customers and financial counterparties to come together more rapidly to put the company on a solid footing.

Subscribe now

Bucket 3: National security, ecosystem resiliency, and other offensive measures

Some of our milestones aimed to transcend the specific project and push outcomes focused on broader national security and resiliency goals. While we wanted each project to succeed individually, the ultimate goal was a robust, interconnected ecosystem with the full range of critical technologies, resilient access to upstream and downstream supply chains, and investment in R&D and talent. Every time a project came to our screening committee, I would always ask the same question: what else could this company do to further our overall goals and build greater resiliency?

Defense and national security were the top priorities for CHIPS. We built an exceptional national security team within our strategy group. The team scored every application based on their deep understanding of specific technical requirements necessary to enhance our security, input from the broad government national security ecosystem, and the project’s potential contribution as described in the application. They would then make specific recommendations on what more the company could do to enhance that score. We converted these recommendations into formal commitments and incremental milestones to strengthen the project’s contribution to national security. Examples included commitments to:

• develop or transfer onshore specific technologies for defense applications;

• continue to produce certain products for longer tail periods and extend notice periods if a technology was going to be phased out;

• take action to make their fab or governance structure more secure so incremental classified engagement could happen with the DOD; and

• strengthen cybersecurity, localized IP, or other operational security metrics to reduce risks of technology leakage.

We also worked to build a more resilient ecosystem. Semiconductor fabs sit within a deeply interconnected global supply chain of materials, chemicals, equipment, components, and packaging. That supply chain is centered in Asia, and largely concentrated in China and Taiwan. For example, most advanced packaging is done at TSMC facilities in Taiwan — how could we build resiliency if any US-manufactured chips would still need to be sent to Taiwan for packaging?

We knew that solving the bottleneck of sufficient US semiconductor fabrication would inevitably expose other chokepoints across the system. Semiconductor manufacturers themselves often play a decisive role in shaping their supply chains; given their power to move their business to alternative companies, they often can require their suppliers and other partners to follow their manufacturing footprint. Our job was to use our funding and milestones to push companies to onshore more of the supply chain or otherwise build resiliency upstream or downstream of their facilities.

Our program requirements often included providing some data and transparency on upstream suppliers. Beyond that, we knew we had to get creative and ask applicants the right questions: Where will you send chips to be packaged? What actions could you take to have that happen here? If we can’t onshore, can you commit to moving a certain percentage of that function outside of China or Taiwan? What would it take to find alternative sources for a Chinese-sourced material? We would then document these commitments and tie them to milestones.

A few other unique milestone approaches come to mind. To manage product availability and inventory, we at times negotiated commitments for company-wide or product-specific days of inventory on-hand (DOH). During the pandemic, inexpensive semiconductors, such as windshield wiper chips, halted entire manufacturing lines because inventory buffers had disappeared — we didn’t want that to happen again.

There were a myriad of other customized commitments that were one-off in nature or negotiated on a case-by-case basis:

• To encourage investment in R&D, we often set a minimum US-based R&D spending or percentage-of-sales floors, requiring R&D to be some minimum percentage of total sales for the company.

• We also set cross-project conditions — for example, some of our funding on one facility could be clawed back if a related project did not happen within a certain timeframe or if the company decided to build a scaled facility outside of the US.

• Some milestones tried to manage idiosyncratic outcomes, like targeted reductions in outsourced chip manufacturing if the project aimed to bring chip manufacturing in-house, or a minimum percentage of total capex to be spent in the US to prioritize domestic development..

Designing broader milestones for these big-picture outcomes was one of the more innovative aspects of CHIPS. We worked to create a connected and sustainable US semiconductor ecosystem for both commercial and defense applications.

Milestones are not just a defensive game

Effective industrial policy is not defined by how much money the government commits, but by how that money is structured, conditioned, and enforced over time. Milestones were the mechanism through which we translated national and economic security objectives into concrete, enforceable actions by private companies. They protected taxpayers against failure, delay, and drift. But more importantly, they allowed us to shape outcomes – to push beyond construction toward operational fabs, viable technologies, committed customers, national security commitments, and a more resilient semiconductor ecosystem. We played offense from the start, challenging companies to stretch beyond their initial proposals.

In a December report on the CHIPS Act, the GAO identified 161 total milestones across the 40 projects that CHIPS awarded. As of July 2025, 24 of those had been completed and Commerce had disbursed $6 billion of funding with incremental disbursement requests being reviewed. Though we’re still in the early days, the evidence suggests that our milestone approach is working.

Public attention focused on headline awards and how much was given to TSMC, Intel, Micron or others — we looked beyond that. We considered what those companies would actually do: when they would build, what they would produce, who they would serve, how they would finance themselves, and how their investments would strengthen US resilience rather than simply add capacity. We set highly negotiated, forward-looking milestones with real consequences. Those details rarely made headlines, but they will determine whether the CHIPS Act ultimately succeeds.

Interested readers can subscribe to ChinaTalk below!

ChinaTalk

Deep coverage of technology, China, and US policy. We feature original analysis alongside interviews with leading thinkers and policymakers.

By Jordan Schneider

We’re here for a CHIPS Act megapod, in person with Mike Schmidt and Todd Fisher, the director and founding CIO of the CHIPS Program Office, respectively.

We discuss…

• The mechanisms behind the success of the CHIPS Act,

• What CHIPS can teach us about other industrial policy challenges, like APIs and rare earths,

• What it takes to build a successful industrial policy implementation team,

• How the fear of “another Solyndra” is holding back US industrial policy,

• Chris Miller’s recent interest in revitalizing America’s chemical industry.

Listen now on your favorite podcast app.

Was the Fab Boom Inevitable?

Jordan Schneider: We’re about a year out now. There was a long arc of Congress and two administrations imagining what the CHIPS Act could be, and now we’re sitting here in the first quarter of 2026 with fabs popping up everywhere — the biggest semiconductor buildout in memory. Everyone in the world wants more chips and more manufacturing capacity, and a decent percentage of that is now coming online in the US. That wasn’t necessarily baked in. Looking back, how do you give credit to the incentives versus the macro trends that would have led to some version of this buildout regardless?

Todd Fisher: It was definitely not baked in. If you go back to when the CHIPS Act passed in August 2022, ChatGPT hadn’t even launched until November of that year. The concept that AI would drive this massive demand cycle was not part of the original calculus — that became clearer as the years went on. At the time, we were projecting a trillion-dollar semiconductor industry perhaps by 2030. Now, that trillion-dollar milestone is going to be hit this year. The demand cycle was not anticipated.

What we did accomplish is acceleration. It’s hard to shift an entire business from outside the US into the US along with all of its supply chains. That takes time, effort, talent development, and construction. The building that’s happened — what we incentivized — has enabled a more rapid participation across the industry.

Mike Schmidt: A massive buildout of fab capacity to meet the moment on AI around the world was inevitable once the AI boom really took off. CHIPS was ultimately about where those fabs would get built and whether we could get a good chunk of them here. The CHIPS Act played a huge role in that.

A few factors stand out. The investment tax credit was enormously important — a 25% cost offset on investment that provided a strong baseline subsidy. Our office managed $39 billion in grants, and we mobilized around that. That worked together with some natural advantages we have as a country: a strong economy, a strong workforce, and — really importantly — the major customers of semiconductors are American companies. When companies look around the world, it makes sense to be close to their customers. A confluence of factors — some market-driven, some about public policy, some about execution — all conspired to put us in the position we’re in.

Jordan Schneider: Taking a step back from CHIPS, when you think about the playbook you ran to promote domestic manufacturing or shape the market, what’s the holistic framework? What aspects of CHIPS worked for and against you as you were trying to put government money behind an industrial aim?

Todd Fisher: Everything’s a trade-off. If you’re going to put — adding the ITC and our subsidies — $100 billion into the semiconductor industry, that’s $100 billion you’re not putting somewhere else. You have to frame it that way: we can’t do everything, so we have to be focused.

I laid out this approach in one of our Substack posts on Factory Settings. Something has to be critical — critical for national defense, for people’s health, for enabling technologies of the future. It also has to be compromised in a way that’s going to cause us pain. We’re seeing that right now with the weaponization of economic tools around rare earths, semiconductors, and other specific choke points that are genuinely negative from a national and economic security perspective. And you have to determine that your effort — your subsidization, your tax incentives — can actually make a difference. That gets to the nature and structure of the industry, the supply-demand dynamics, and the cost factors, and whether a subsidy over some period of time can shift the fundamental economics.

Those conditions won’t be true across the board for all industries, so you need to evaluate all of them when designing industrial policy.

For chips, it’s clearly critical — they’re an enabling technology for just about everything happening right now. You can look at the newspaper articles every day about memory shortages and the need for TSMC to make more chips. That is going to be the ultimate chokehold on AI. It’s clearly compromised, because up until a couple of years ago, we had zero percent of leading-edge logic or memory being produced in this country. Now, for the first time in over a decade, we do. And in our view, it’s changeable, because the semiconductor industry is not labor-intensive — it’s highly automated. The cost differentials are much more focused on front-end construction than back-end operations. If you can correct the cost differentials on the front end, there’s a path to sustainability over time. Those are the aspects that made sense for leading-edge semiconductors specifically, and that framework can be applied elsewhere.

Mike Schmidt: The big analytical project of the moment is figuring out which industries have a similarly strong and compelling national security case for intervention. Once you decide intervention is necessary, our experience with CHIPS suggests pretty strongly that you have to study that industry closely to understand the tools and approaches that will work — whether that means onshoring, working with allies and partners, or otherwise creating a more resilient supply chain.

Jordan Schneider: Let’s talk about the sequencing. We have this bill, and in it there’s a little bit for advanced packaging, some money for R&D, but it’s basically telling the executive branch, “Figure it out. We think this is important and we came to a number. You go from there.” That seems like the wrong sequencing to me — I don’t know.

Mike Schmidt: You need a mix of purpose, direction, and discretion. There were two sides of CHIPS — $11 billion for R&D and $39 billion for manufacturing incentives. Our part was the $39 billion for manufacturing.

Jordan Schneider: Right, we don’t talk about that other $11 billion.

Mike Schmidt: For us, it was basically a blank slate. There was one $2 billion carve-out for fairly mature technologies, but we wanted to spend more than $2 billion on that anyway, so it wasn’t a binding constraint — it was a floor rather than a ceiling. That put us in a position of asking, “What do we want to achieve with these funds, and what do we think we can achieve?”

Early on in program implementation, when you’re just building a team and racing against the clock to get your funding announcement out the door so you can take in applications, there’s a baked-in narrative that things are moving too slowly. It took a fair amount of discipline for us to say in those early days that we needed to invest the time and resources to really articulate a vision.

We ended up doing that through a document we called our Vision for Success — Dan Kim was on a couple of weeks ago and talked about it as well. That was our stake in the ground. We had a lot of discretion, and we wanted to hold ourselves accountable. We did a whole bunch of work to figure out what our objectives would be, and that document ended up being a really important disciplining mechanism — not just externally but internally — because it created the framework by which we would measure our own success.

Todd Fisher: You said you’re not sure it was the right sequencing. I actually think it was. Identifying the big issue that could potentially move the needle for the country — semiconductors broadly — and then enabling the people who would actually execute and implement the program to dig in with the right expertise and think through different approaches: that’s the right way to do it.

If you legislate something with too much detail, you’re going to miss the nuances of execution and implementation that are so important. The effort we put into the Vision for Success, the way we thought through program design and incentives, the kinds of companies we wanted to approach, who we wanted to draw in, how we set up the whole process — we were quite lucky to have the flexibility to design it from the ground up.

Mike Schmidt: That said, you can go too far in the direction of discretion. For us, we knew it was chips. You could imagine a world in which, given all of these vulnerabilities and choke points, you take a huge pot of money and just give it to the executive branch with the instruction to advance economic and national security. There are benefits to that — you can imagine a robust, flexible executive making dynamic decisions, addressing problems as they emerge, and moving quickly.

But with CHIPS, we had a huge benefit from the fact that we had a bipartisan law with a clear overarching objective, and we could then define the specifics within that. It allowed us to build a team centered on that objective, develop deep expertise in the industry we were focused on, and build internal know-how. In the broader political context, it also gave us a clear measuring stick that we’d be held against.

Jordan Schneider: The backdrop of what I was getting at is that the origins and legislation around the CHIPS Act were this random confluence of an intellectual effort plus COVID, and it happened to be the right moment with enough momentum behind it. If you ran 2018 through 2022 thirty times, the CHIPS Act probably happens maybe two or three times. But the question is, what if you had $85 billion to do American economic security broadly? Would we be more economically secure today if that had been the mandate, as opposed to directing all the money specifically toward chips?

Mike Schmidt: Personally, there’s a lot of value in Congress identifying key target areas and then telling the executive branch it has broad discretion to figure out the how. You can spin a lot of wheels trying to identify targets from scratch.

Jordan Schneider: The IRA is a counterexample, right?

Mike Schmidt: The IRA, unlike CHIPS, isn’t one pot of money or a single tax credit doing one thing. It’s an assortment of different tax credits and programs, each with a different objective. A challenge with the IRA is that some of those objectives are targeted at different things, and how they all work together into a coherent strategy is tougher than when you have one specific goal and a few tools to mobilize toward it.

Todd Fisher: But why do you say the IRA is a counterexample?

Jordan Schneider: Maybe it’s not. The real counterexample is that no one would ever get $100 billion to do economic security in the abstract. The happy medium — where the executive branch has enough trust from Congress to actually do things that don’t get earmarked away into, I wouldn’t say oblivion, but not the strategic version of the thing — probably requires Congress signing off on a particular industry.

Todd Fisher: I probably haven’t thought enough about this, but the CHIPS Act for an industry that’s going to be a trillion-dollar market was very appropriate. Take rare earths and critical minerals today. It’s hard to just wait for Congress to figure out this is an issue and pass legislation. There’s some combination needed. The right answer isn’t “here’s $100 billion, figure out what to do with it.” It’s more like: chips are important, we’re going to do something very focused there, and here’s $10 billion to address things that are going to become critical.

Consider the rare earth industry. These are 17 different elements, and the overall market is about $5 to $6 billion. You can solve the rare earth issue because it’s a $5 to $6 billion problem. You can’t solve a trillion-dollar industry the same way. There’s a different mindset required. You could look at APIs, pharmaceuticals — there are different choke points across the system that we need to be more nimble about addressing. Then there are the big things that take massive amounts of time, effort, and money.

Jordan Schneider: The takeaway is that there are two buckets — giant industries where the amount of money needed requires direct congressional appropriation, and then a second category where it would be nice to have an office that looks at these issues with a smaller pot of money and the ability to issue debt or whatever other tools — things that are an order of magnitude smaller where we don’t have to stress as much.

Mike Schmidt: That’s a good idea. I like it.

Jordan Schneider: I stole it from you.

Mike Schmidt: There’s also a question of what kind of state capacity you need to develop. For something like pharmaceutical APIs, it might make sense to have a dedicated program because it’s such a specialized market requiring deep technical expertise. Whereas something like critical minerals might be more straightforward in terms of what you’re trying to achieve and how the industry is structured.

Jordan Schneider: When it comes to this idea of discretion and Congress trusting you, we’re in a different world today. I remember you being very focused on transparency — wanting to make the process fair and making sure everyone felt they were getting a fair shake. Now we’re living in a world where relatives of presidents and cabinet members sit on boards of companies receiving government money.

It feels almost quaint in retrospect how focused you were on that. But my worry with the smaller, more nimble and discretionary efforts is that we’re going to enter a backlash to the current moment where there just won’t be any appetite or trust for this kind of thing.

Todd Fisher: That all comes down to governance and how you set things up — specifically, the ability to insulate the process from politics and preferences that intervene in unconstructive ways. The good news about CHIPS is that it was set up in a way that our team felt pretty protected from politics. That was partly structural, partly how Gina Raimondo approached it, and partly about personnel. Neither of us were political appointees. Many of the people I hired on my team were Republicans. It depends on how you set these things up — whether it’s purely transactional wheeling and dealing, which is one approach, or whether you build something more institutional. You could set something up along the lines of the DFC, where it’s insulated and more trustworthy, with built-in transparency. You’re right that there might be a backlash, but I would caution against retreating from this, because we’ll miss something really important if we do.

Jordan Schneider: Could you do the short version of that, Mike?

Mike Schmidt: On balance, I would urge simple tax credits and other straightforward tools as the default. There’s probably an additional bar you have to clear before deciding when the discretionary approach makes sense, because the potential for politicization and abuse is real.

For us, the semiconductor industry was such a distinctive case. The scale was dramatic, the industry was highly concentrated, and creating a hub in the government that engaged dynamically with the industry to form partnerships, scale up investment, and make specific commitments we felt were important for national security — all of that proved really valuable.

There are other circumstances where discretion can be helpful. On rare earths, discretion right now is probably useful just because it’s such an emergency that it doesn’t make sense to throw a tax credit out there and see what happens. But that’s partly because our own policies created the emergency. A longer-term, more stable industrial strategy would identify a set of priorities, establish a baseline of market-based incentives and tax credits, and then layer in discretion for the key areas where it really makes sense.

Todd Fisher: I broadly agree, but the challenge is that tax credits only go so far — they’re really just one tool. When you start thinking about rare earths, some of the things the Trump administration has done are very credible, particularly on the demand side of the equation, providing price floors and a certain amount of guaranteed demand.

We didn’t have that. One thing we really lacked in the CHIPS Act was any kind of demand incentive. We were forced to rely entirely on the bully pulpit — engaging with Apple, Nvidia, AMD, Broadcom, and others, telling them this was in their interest and that they needed to be helping us so we could help them drive supply development. Aggressive supply expansion only happens with real demand signals. Those exist today, but they didn’t four years ago.

I prefer the tax credit any day, but it’s a very evenly spread incentive, and you miss a lot. You miss a number of tools, and you miss the ability — which we really took to another level — to push companies to do more. With a tax credit, companies are going to do what they’re going to do. With our discretionary money, we were able to say: no, we need you to build another fab. We need you to bring this technology here. We need you to shift your inventory levels. That allowed us to be much more proactive about our priorities — like drawing advanced packaging facilities to the US, which wasn’t on the table for the first year or so.

Mike Schmidt: The other thing, from an institutional standpoint: Treasury’s Office of Tax Policy is a fantastic institution within the government. I worked very closely with them in my role before CHIPS. But they’re never going to be the hub of energy in government to make an industry succeed — that’s not their role. They write regulations to make sure congressional objectives are being met, that credits aren’t ripe for abuse, that there aren’t loopholes. That’s not what we did. What we did was build a group of people — 180 at its peak — who woke up every day and thought: how can we make the semiconductor industry successful in the United States, and what tools do we have at our disposal to do that?

Jordan Schneider: There’s so much to pick up on. I’m very thankful that your child was two years old and not a 25-year-old hotshot semiconductor hedge fund member.

The institutional question is a really interesting one, though. You alluded to the DFC. The Fed is another example — though I guess we’re politicizing that one too now. But the idea that you could have some chunk of the government whose job it is to prevent China from having economic leverage over the US, or to do everything possible to minimize that — with a dedicated pot of money and tools over a sustained period — seems like a bipartisan thing.

Mike Schmidt: There’s a lot to disentangle in that. Having some chunk of the government that focuses on this is really important. Even the Trump administration ended up taking on the CHIPS office and putting it inside what they called the United States Investment Accelerator. There is this notion that what we built could be the foundation for something more durable — not just about semiconductors, but broader. A permanent institution that pulls people from the private sector and public sector to focus on this set of problems, one that develops methodologies, rigor, and expertise around it, would be hugely valuable.

Jordan Schneider: We have the U.S. Digital Service with these little rotational programs for software engineers. Why not industry analysts?

Mike Schmidt: It seems like a natural fit. And it doesn’t have to be rotational — it could just be that they’re in charge of allocating whatever pool of money exists. There are two downstream questions worth pulling apart, though. One is how much discretion this office has over what industries receive funding. Does Congress set the objectives and identify the priority industries, then hand it off to the office? Or does the office have more discretion? The second question is about independence from political interference and finding the right balance. We created systems and structures within our operation to keep politics out of it to the extent we could, but that wasn’t legislated. It was about establishing processes and norms. There’s a whole set of questions around how to formalize that.

Todd Fisher: I go back and forth on this. The special nature of what we had was the ability to attract the talent we attracted. Most of those people have now left — not only because the administration shifted, but because most of them weren’t necessarily partisan people.

Jordan Schneider: They couldn’t do math fast enough?

Todd Fisher: No, but obviously they should have had better talent. Come on.

They came from the private sector and wanted to go back to the private sector. The rotational aspect you mentioned would be great. But the thing that makes me nervous about setting up a separate institution — and on balance, I do think it would be beneficial — is how you sustain a pipeline of really talented private-sector people. Not only financial and investment types, but semiconductor and industry experts. How do you bring them in dynamically, in a way where people want to be there and feel that the experience furthers their career rather than setting them back?

There aren’t many examples in government that pull this off. The Fed is a good one, but it’s very difficult to set up another Fed. That’s going to be really hard. Then you say, well, maybe the right answer is that when you have something big, you just do it again and figure it out. I’d love for there to be a centralized function, but I have doubts about whether it would be excellent or merely good.

Mike Schmidt: The counter to that — and my sense is that, yes, a lot of people we worked with left, but they’re hiring people now, and many of them are pretty good. The counter is that this set of issues is going to be at the forefront of national security and competition with China for the foreseeable future. That naturally draws people. People came to us not just because they thought we were awesome or because they loved Gina Raimondo — though that was part of it. A big part of it was simply that people want to serve. That piece will continue.

Having served around ten years in government, there’s this enduring mystery — why do certain agencies thrive and maintain an awesome culture over time, while others just atrophy?

Todd Fisher: The other thing is the fragmentation. One of the things that would be great to fix is how many different entities are involved. Even for us, we had to deal with Treasury on the tax side, and even within Commerce, export controls were separate. Then when you start talking about demand and scale, you’ve got the DOD, the Department of Energy — and even at the White House level, how do you create something that integrates all of this?

Some countries have models worth studying. Japan, for instance, has a structure more focused purely on economic security that combines all these functions and breaks down those barriers. It would be really valuable to have some of those barriers eliminated and bring together cross-functional disciplines.

Jordan Schneider: There are two pieces of this. One is mapping out the Chinese economic escalation ladder — the universe of disruptions that could devastate the American economy. Map all of that out. Develop playbooks with dollar amounts, GDP estimates, and labor projections for every contingency we could face over the next five to ten years. Then you can present the executive branch or Congress with a well-thought-out menu: here are the problems, here’s what it would cost to fix them, here are the five tools available. And this analysis is independent — not coming from the chemicals industry, a rare earth miner, or a semiconductor manufacturing firm, but from something that’s working on behalf of the American people.

The question is whether that analytical work alone is exciting enough to attract the best and the brightest, or whether you also need the ability to say, “We’re solving this, and we have real money to deploy.”

Mike Schmidt: You need the money side for the best and brightest.

Todd Fisher: One hundred percent.

Mike Schmidt: The analytical side already exists to some degree — ITA at Commerce has a lot of that capability and a lot of great people.

Jordan Schneider: But they had two people on it.

Mike Schmidt: Right. You could definitely imagine a more robust version of that. But to truly attract a broad range of people who want to make a difference, you need more than just analysis.

Jordan Schneider: People actually want to do something.

Mike Schmidt: Exactly. Telling someone, “Come up with a plan that Congress may or may not enact” — in an era when Congress doesn’t seem to do very much — it’s hard to see that being enough. One of the most humbling experiences of doing CHIPS was the commitment to the mission. People would come in and just work their asses off. They stayed longer than they planned. They really wanted to get the job done and felt it was important. A lot of people stayed well into the Trump administration because they were focused on what mattered. That’s always awesome — truly cool to see. But people show up when there’s really something to do.

Jordan Schneider: Right, but that’s the argument for having a small sovereign wealth fund-type entity where you can pitch to an investment committee or something.

Todd Fisher: You have to have capital.

Mike Schmidt: If money is there but the question is how it’s going to be deployed, that’s a big job — a very different thing. You need the money to build state capacity. It falls to those of us outside of government to figure out the right framework for Congress to act on in a bipartisan way.

Todd Fisher: Don’t call it a sovereign wealth fund. A sovereign wealth fund, in my mind, takes surpluses that a country has and invests them to enhance returns for the government. This is more of an investment entity — I don’t know what the right term would be — where you’re actually trying to influence industries in a more targeted way.

The concept is great. What I struggle with most — and Mike as well — is that what we did over those two and a half years is hard to look back on and say, “That’s repeatable.” It’s not easy to replicate. You can’t just hand someone a playbook and say, “Go execute.” It feels like a one-off. And I believe strongly that for us as a country to be successful long-term, it can’t be a one-off — it has to be repeatable. I want to really believe we can create a separate entity that attracts the best and the brightest so we can do it again. That’s something I struggle with a lot.

Jordan Schneider: This is why I want to come back to the corruption piece. Part of what made this a one-off was that the team was above reproach. The pressures you faced were none of the really ugly ones we’ve seen throughout American history when the government spends lots of money. What worries me is that what we’ve seen over the past year — and will continue to see over the next three years — is going to poison the well for this kind of work. Work that all of us agree needs to happen. People just won’t trust it.

Mike Schmidt: There are many respects in which what’s happening in the world right now is going to make this kind of work harder going forward, and that’s certainly one of them. If you begin to question the integrity of these efforts, it makes it harder to sustain the political will — or you end up choosing different tools, like tax credits. That’s a big piece of it.

But there are others. The whole discussion about allies — we spoke throughout the Biden administration about the importance of working in concert with allies and partners. There’s a moment right now, in the wake of Carney’s speech and this broader reckoning with what’s happening with our traditional allies and partners, where thinking through new frameworks to create durable strategic economic connectivity is really important. There’s probably a longer list than that, but it’s a critical issue to raise.

Jordan Schneider: It’s frustrating to watch the critical minerals work, because clearly there are really good people doing really smart things — that’s ninety percent of it. Then you get a story or two and you think, “What the hell?” Stopping a negative outcome isn’t going to resonate with the American people. Having a critical minerals reserve isn’t going to get headlines or clicks. What gets attention is someone controversial ending up on a board.

Mike Schmidt: But there’s a more hopeful flip side. The folks doing this work now, in the trenches figuring out which tools to apply to which situations, running into the same state capacity constraints — or maybe blowing through them in ways we didn’t — over time, they could develop into a group of capable technocrats who inform this work constructively going forward.

Todd Fisher: I’m actually more positive on the critical minerals and rare earths front. There are multiple things happening. Getting stuff done in government is hard, for all the reasons we just discussed. But being able to break through logjams, push things forward, and deploy multiple tools — that’s what’s happening in the rare earths space. Multiple efforts are underway, multiple tools are being employed across the board. I may not agree with all of them, but broadly speaking, deploying those tools aggressively, building a strategic reserve, and bringing countries together to think through the issue collectively — those things fit together. You’re right that one negative can erase all that good work.

But hopefully this is a sample of what you might be able to do with a smaller industry, because it’s a solvable problem. It takes time, but it’s totally solvable in dollar terms — even if we had to subsidize it forever, given the scale involved. That would not be true of many other areas where there are real choke points.

Share

Jordan Schneider: Let’s talk about doing things that lawyers get upset about. Mike, how hemmed in did you feel? If you knew then what you know now about how far an administration could push things without major blowback, what would you have done differently?

Mike Schmidt: We tried to do as good a job as we could thinking about risk holistically, and that’s true at multiple levels. With individual deals — Intel is a risky deal, but there’s a lot of risk in not doing a deal with Intel, given the critical role they play in the broader ecosystem. But it’s also true when it comes to questions of process.

One of the things I used to always tell the team as we were thinking through how to get something done: the biggest risk of this whole thing is that we don’t get it done. We’re going to run out of clock, we won’t have gotten our deals across the finish line, and the country will be worse off for it. We needed to think flexibly and dynamically about all of this. We did a pretty good job of that, but looking back, there are areas where it’s really worth asking: could we have been more willing to take risks? Could we have prioritized efficiency over some of the risk avoidance?

Todd Fisher: Where would you have taken more risk?

Mike Schmidt: We have a new Substack where we’re talking about some of this. We discussed how hard it was to get our deals done. For listeners, we would reach term sheets with our applicants laying out the basic economic terms — how much money, over what milestones, and so on. Then we had to turn those term sheets into final award documentation, which meant diving into a bunch of nitty-gritty legal details.

Each individual issue involved a real trade-off, and you could understand why the government cared about it. If the government gets sued because of something the company did, should the government be protected? If the company tries to sell a project or sell itself, should the government have the ability to reassess the deal? If the company breaks the law, violates export controls, violates sanctions — all of these provisions accumulate to create a huge amount of friction in getting deals done.

We thought about each one rigorously and made a lot of trade-offs along the way. It’s not as though we weren’t trying to be flexible. But when you talk about sustainability, I look at that and say I would definitely advise a team going in to do this again to put all of that on a list and just decide up front: we don’t care about all of these. Cut twenty-five or fifty percent of them. That means accepting more structural risk in the deals, but it makes the whole process much more manageable and dollar-efficient.

Todd Fisher: I agree with that. What keeps coming to mind for me is the demand side of the equation — how we could have pushed Nvidia, AMD, Apple, Broadcom, Qualcomm, and others harder to help accelerate things and set up an environment that moved faster, particularly with some of TSMC’s competitors. Whether we could have done more, I’m not sure, because we didn’t have the actual tools. It was mostly the bully pulpit.

Jordan Schneider: Was there ever a thought to go back to Congress on that side?

Todd Fisher: I don’t recall. Getting new appropriations for demand incentives — particularly giving money to some of the largest trillion-dollar companies in the world — is a difficult ask. We structured some things where we agreed to pay for porting costs to incentivize companies to dual-source or shift some of their work from TSMC to Intel or elsewhere. But it was all cajoling. Could we have pushed harder? I don’t know. We didn’t have the tools, so it was just persuasion.

This administration is bringing every tool to bear in a very transactional way — “if you don’t do this, I won’t do that.” That wouldn’t have been an approach we’d have been comfortable with. But between where we were and where things are now, there’s probably quite a bit of space in between.

Jordan Schneider: Instead of asking for daycare, threatening to get CEOs fired would have been an interesting approach.

Mike Schmidt: It is a question of norms — how you interact and interface with industry. It would have been tempting. I remember at one point we had a conversation about a company we were dealing with that we knew had something going on with BIS. You could cross those two things pretty easily and make it much simpler to extract the investment you were looking for. But there was just this sense — I believe it was Leslie, our general counsel, who said, “We don’t do that.” And the rest of us were sitting there thinking, “That seems right.” Because once you start going down the path of the government exercising its power in one area to get what it wants in another...

Jordan Schneider: Then you get donations for the East Wing. That’s how it all comes together, right?

Mike Schmidt: Right. That felt like a line. But at the same time, maybe we were a little too precious about it. That’s an interesting question in terms of finding the right balance, because at the end of the day, you’re trying to get these companies to do things that are important for national security.

Todd Fisher: That’s the one area where people might push back. You could say we should have tried to get TSMC to build more fabs, but to me that’s a total red herring. On TSMC, our major effort was to get them to build three fabs. Why three? Because of the way they construct their fabs, once you build the third, it’s almost guaranteed they’re going to build a fourth. Once you have four fabs, you’re going to build six because you want that mega-fab scale. Our view was that we didn’t want to spend the extra money to get another fab up front. If we could get them to three, the rest would take care of itself.

The fact that they’re now saying they’re doing more — because demand projections moved from a trillion dollars in 2030 to a trillion dollars in 2026 — doesn’t really mean anything to me. It’s not as though I anticipated that acceleration. The other question you could ask is whether we could have pushed these companies harder to do more. We pushed them really hard, and we got them to do a lot. There’s the public stuff — the big numbers and the specific fabs. Then there are things in the background that people don’t focus on or don’t even know about: moving certain technologies here, agreeing to provisions in their deals around supply chain, upstream, and downstream that will strengthen the ecosystem very significantly. I feel very good about that.

Jordan Schneider: If we’re going to use TSMC as an example, there is a world in which a president, instead of being happy with a $500 billion number on a piece of paper, says, “If these fabs aren’t built by this date, you’re not getting this.”

Todd Fisher: That, and more.

Mike Schmidt: We did have that — with our money. That was the milestones. But it was within the bounds of the program, as opposed to saying, “If they’re not built by this date, then some licenses get compromised” or other levers outside the program.

Jordan Schneider: It comes back to the independence-versus-leverage question. You say you were pushing them pretty hard. You were mentioning World War II industrial policy analogies earlier. If a war was starting, or if you had ninety percent confidence it was going to start in two years and you needed TSMC to move half of its company to America in six months, there are levers the U.S. government could pull that are very different from tax incentives.

Mike Schmidt: True. Although, from my amateur historian reading on how World War II played out, it was a full mobilization with a huge buildup of state capacity. But the tools were pretty straightforward — the power of the government as a customer and as a financier on the supply side. In the meantime, there was massive political mobilization on the industry side — to win the political narrative coming out of the war and lay the foundation for the type of post-war economic system that the business community wanted.

You don’t read that history and get the sense of executives who were afraid the government would use leverage in one area to compromise their interests in another. You get the sense of the government having its interests, industry having its interests, and ultimately both sides working incredibly collaboratively to get it done — but with a huge amount of friction, tension, and politicking along the way.

Jordan Schneider: What I’m getting at is the dynamic with allied countries that are small, that can be pressured, where the U.S. has an enormous amount of leverage. You were talking earlier about not coloring outside the lines — not crossing streams with colleagues at BIS, not stretching the boundaries of executive branch authority. One of the lessons of the Trump administration over the past year is that the executive branch is incredibly powerful, and the amount of leverage America has over other countries is remarkable.

Maybe it’s better to keep that Pandora’s box closed, because opening it leads to the really dark outcomes we’re seeing now. But I can’t help feeling that however much success the CHIPS Act has had, the current administration is striking the fear of God into these companies in a way that nice subsidies never did. A different dynamic could have been unlocked.

Mike Schmidt: It’s a worthwhile discussion — the intersection of diplomatic and industrial activities and how those two things interact. Our experience was that we had our own lanes with industry. These companies are massive power centers. If you look at the major Korean or Taiwanese chip companies, within their own countries they are not the government — they have their own interests and their own geopolitical positioning.

But in the backdrop of all that, the strategic relationships between countries were always present. For us, those were mostly separate threads, but they were generally swimming in the same direction. As we were doing our Samsung deal, Biden was holding his summit at Camp David with Korea and Japan, creating new strategic dynamics in Asia that were pulling allies closer. We always felt our work fit into that, and I’m sure the Koreans felt the same. We had our lane — dealing with Samsung and SK hynix to secure the investments we needed — and it was part of a bigger picture.

Jordan Schneider: Maybe it’s a question of the difference between looming crisis trend lines we don’t like, versus having six months to actually prepare for something that the president says is important.

Mike Schmidt: But as a political question, the real challenge is — how do you create a sense of crisis before it happens, so that you’re actually prepared?

Subscribe now

Defensive vs. Offensive Industrial Policy

Jordan Schneider: You said on another show, Todd, that you did ten times the deals KKR would do in a year in about eighteen months. What do you lose at that pace? What do you gain? What are the trade-offs?

Todd Fisher: You lose sleep, honestly. When you’re doing deals at KKR, you’re out hustling to find the right opportunities. Here, the deals were coming to us, so it’s not a perfect comparison. The more relevant parallel is that once you have an agreed deal, getting it to a final, documented, sign-on-the-bottom-line agreement is hard. Mike just talked about that. It’s also true in the commercial world, but there the boundaries are already set. KKR has done hundreds of deals in its history and has hundreds of agreements with companies. There’s only so much room to go back and forth. That was a real challenge for us — trying to get multiple deals done in a very short period of time.

But it also creates excitement, team dynamic, and a culture of, “We’re just going to do this and get it done.” That’s very energizing. Having everything on the table at once, with everybody working and trying to get it all across the line — that was very positive.

Mike Schmidt: There’s also a question underneath your question. We often talk about process as something that gets in the way of the objective. But to what extent does process, even if it takes more time, actually end up serving the objective?

One thing we’ve hashed over in retrospect — and a little bit at the time, but mostly afterward — is this wonky structural detail. Our CHIPS money functioned mostly as grants. It’s basically free money that companies keep to help offset cost disadvantages relative to Asia so they invest here. But legally, it wasn’t grants — it was under something called Other Transaction Authority. That meant we had much more flexibility than we would have had in a grant context to design the program as we saw fit.

One thing we could have done is come out of the gates and start doing deals immediately. Instead, we put out a Notice of Funding Opportunity (NOFO), created structured evaluation criteria, built an application process with timelines, and built the team around that. That meant we weren’t immediately going to Intel and TSMC saying, “All right, what’s the deal going to be?”

We also used that time to build our team and develop capacity. But there’s no question: the decision to create that structured process cost us time. At the same time, it had a lot of benefits. It put us on footing with industry where companies knew how we were thinking. They knew our objectives from our vision for success. It grounded negotiations in a sense of fairness and rigor, and ultimately built mutual credibility and respect. It certainly provided some political protection — we laid out a process and followed through on exactly what we said we’d do.

All of that is good, but it takes more time. There are people on the team who would think I’m crazy — they’d say we should have just been out there doing deals faster. I get it. But our sense is that it was probably good to bake in some process to slow our roll a bit so we didn’t rush out and do something we’d regret later.

Todd Fisher: That’s right. I believe strongly in that. If your goal is just to get a deal done with TSMC and Intel, then just go get a deal done with TSMC and Intel. But our goal was broader than that.

Advanced packaging is a perfect example. If we hadn’t laid out a process, really gone after that space, and recognized the critical issues in advanced packaging, we wouldn’t have gotten Amkor to invest here. SK would have been a question mark. Those deals are going to be critical to the future.

We haven’t talked about this, and we’re not as knowledgeable on the R&D side. But one of the things on my mind is that right now, as a country, we’re focused on what I’d call defensive choke points — “Oh my God, there’s an issue in critical minerals, we’ve got to address it,” or “We have zero percent bleeding-edge semiconductor manufacturing here.” The next one will be batteries, or drones, or APIs, or whatever it might be.

But what we should also be focused on are the enabling technologies of the future. Advanced packaging is one of them. The industry is changing dramatically. TSMC broadly owns it now, but it’s evolving, and the U.S. can become the major hub for what is essentially the next generation of Moore’s Law. The same is true across many different materials, substrates, and areas where we should be investing more on the R&D front in a more offensive way.

A lot of what we’ve talked about today is how to use industrial policy to plug choke points. But it’s also about how to use industrial policy to think about what will enable the future. Is it going to be humanoid robots? That full supply chain basically doesn’t exist today but could potentially be created in the U.S. and become a fundamental competitive advantage over time. But we’d have to focus on it and figure out what it’s going to take to get there.

Jordan Schneider: The offensive versus defensive split is interesting. On the defensive side, I feel confident that if you set up a Fed-like institution, give them $50 billion, they’ll 80/20 their way there — these are manageable, tangible problems. The money can be directed to solve specific issues, and the technology already exists. But the offensive stuff is different. There are pieces America has that we’d presumably want to keep — global dollarization, TSMC and Intel and Samsung being ahead of where SMIC is. But beyond that, I’m really not sure I trust the government to pick what the next industry is going to be. The answer is probably more that the NSF and NIH need to do their jobs well. It’s a much more amorphous challenge of how to get R&D right.

Mike Schmidt: It is more amorphous and definitely extends beyond our direct experience. But when you start thinking about the offensive side, you also need to expand beyond what you might traditionally think of as industrial policy — or expand your concept of what industrial policy is. Because it’s about technological leadership, dollar dominance, our university system, allies and partners. It’s a broader geopolitical game where we have pretty significant vulnerabilities that have emerged vis-à-vis China. They’re going to continue investing, continue looking for ways to create more vulnerabilities or shore up their own.

Todd Fisher: We have to do the same. Those vulnerabilities emerged because China took a more offensive approach. The fact that they identified the electric vehicle supply chain as critical — particularly the enabling technology for EVs like batteries, which are also essential for drones, robotics, and more — that positioned them well. What I’m saying is there’s a role for government to incentivize and encourage some of this activity. Obviously there’s a lot of venture capital. We have so many great advantages in this country — deep financial markets, deep venture capital, innovation, phenomenal design capability. But the government has a role to play in complementing all of that.

Mike Schmidt: We are just so far behind on the manufacturing side that when I think about manufacturing activity, I immediately go on the defensive. Maybe in that context, there are some areas where we can carve out our own sources of leverage. That’s why my head ends up going toward these other dimensions — it’s not just about manufacturing. But I agree with what you’re saying.

Jordan Schneider: This is maybe the fundamental question of the 21st century — can you R&D your way out of needing manufacturing scale? That’s very much an open question.

Todd Fisher: Why can’t you R&D your way out of manufacturing scale? The whole AI revolution is going to disrupt so many industries, including manufacturing, and we should be — and are — trying to take advantage of that. We have a lot of the skill set to do it.

Take printed circuit boards as an example. We couldn’t fund them through the CHIPS Act because they’re not semiconductors. But if you look downstream from the chip, every chip has to land on a printed circuit board. At least sixty percent of those boards are manufactured in China. As AI has gotten deeper and more compelling, the complexity of those boards — the multi-layered nature of them — has become more important, and they could become a choke point in data centers.

The challenge is that right now, it’s almost impossible to figure out how to meet those costs competitively in the U.S., even with reasonable subsidies. But can you redesign the manufacturing process? Can you not only automate but really rethink it? Are there other tools? That kind of innovation can take a very significant amount of cost out. If we’re able to figure some of those things out, we can be much more competitive in manufacturing at scale.

Subscribe now

Risk Aversion and the Solyndra Mentality

Jordan Schneider: I get pitches in my inbox all the time from startups trying to solve the rare earth problem with some new manufacturing technology so we don’t have to dig the next hole in the ground.

Mike Schmidt: Sounds like you have an interesting inbox. I take it they want to come on the show?

Jordan Schneider: They want Congress to notice them. It’s great. But the question is, if you’re sitting in your American Resiliency Fund, how much money are you spending on these breakthrough technologies versus the mine in Montana? You’d have to make an expected value calculation. What would history tell us? We don’t need guano anymore.

Maybe if demand gets tight enough, sufficient scientists, money, and energy will converge to solve these problems. But the issue is there isn’t money for this stuff because it’s so cheap.

Mike Schmidt: A couple of things on that. One is creating incentives or structures on the demand side so that demand isn’t just being driven to China — so it’s naturally being pushed toward, if not the United States alone, a broader ecosystem of countries outside of China.

In terms of what we do, if we’re going to pursue this on a broader scale — and it’s foundational to our competition with China — we have to get comfortable with the notion that not every bet is going to pay off. That’s true at the company level, and it could also be true at the industry level. Rare earths are a crisis now. We should do those deals. But should we also be doing deals that make those deals look bad in retrospect? Yes! We should be pursuing bets that leapfrog whatever the current technology is.

That’s okay. It’s okay to have some bets pay off and others not, as long as overall we are orienting ourselves strategically to deal with a really significant challenge to our power.

Todd Fisher: That’s not an easy mindset and culture to shift. The whole Solyndra mentality — I don’t know how many times we heard that term. It was constant.

“We can’t have another Solyndra” became code for being really risk-averse. But government money exists precisely because the private market can’t fill that gap, which means you’re inherently taking on more risk than private markets would. Some things are going to fail, and we need to figure out how to build a government culture and oversight approach that’s comfortable with that.

I want to ban the word “Solyndra from government.” It drives risk aversion and prevents people from taking smart bets. The same Loan Programs Office that funded Solyndra also funded Tesla. On an overall basis, that fund did fine — and if you had included some upside sharing or equity, it would have done phenomenally.

Jordan Schneider: I’m just trying to think — would I rather have had Tesla go bust?

I wanted to come back to the CHIPS Act dealmaking side. You mentioned AI tooling. This seems like something that’s very hard to automate. What percentage of the work do you think could have been automated?

Mike Schmidt: Very little. Just to give a sense of scale — I’ve been reading books about World War II, and you’ll come across a passage like, “There was this big challenge, so they spun up a bureaucracy of 25,000 people to solve it.” Our team was 180 people managing $39 billion. A pretty lean organization.

Once you get the deals done, there’s a whole post-award phase — the monitoring piece. We only had maybe a couple of months of post-award experience after getting deals closed and cash out the door. But that ongoing monitoring feels like an area where there might be some opportunity for automation. Maybe in due diligence as well.

Jordan Schneider: There’s a whole universe of investment banking work where the percentage of the junior banker’s job that AI can handle is growing over time. But the bigger point is that it feels like there’s something fundamentally human about the way you set things up — where you actually need to deeply understand the individuals on the other side of the table representing the companies.

Todd Fisher: Ultimately — and maybe this is what’s being compromised right now — it’s about building trust between the private sector and government. What companies were nervous about was: “We trust you, but we don’t know who’s going to be in your seat in a year, let alone five or ten years.” If we’re going to do industrial policy well, the private sector and the public sector need to trust each other in the way Mike described from World War II — people collaborating in trusting relationships while still being competitive and pursuing their own interests. We created that dynamic, and you can’t build it on anything other than a human level.

Mike Schmidt: There’s no replacement for that. Although I will say, I made some really bad slide decks that would probably look great now with AI. Should I pull these out? I was just a big text-on-the-page guy.

Todd Fisher: Well, Mike was better at it than I was.

Jordan Schneider: My wife refuses to download Claude for Excel. I was trying to sell her on it — “Don’t you want to just try it out?” Her argument was, “If I don’t build the model myself, I don’t know what the numbers are.” Working backward doesn’t force you to do the thinking — to actually put the assumptions in your brain down on paper.

Mike Schmidt: That’s a fair point. Todd, when we got our jobs, the first few months — in addition to putting together the vision for success, thinking through how the process would go, designing the application, all the work on the NOFO, building the team — we did a lot of engagement. We probably had 15 different calls with people in the finance industry: bankers, private equity folks, equity analysts, just asking about the industry. The cool thing about these jobs is everyone will talk to you.

Then came our early engagements with the companies and the customers. It was a crash course — just absorbing as much as you possibly can. There’s no way to replicate that through reading alone. At least for me, it’s about hearing things over time and then developing little intuitions about what’s going to work, what’s not going to work, and what’s important.

Todd Fisher: And developing those relationships. I would go to investor conferences and do one-on-ones — 15 investors in the course of a day — just hearing their views on the industry and what they thought was going to happen, while giving them a sense of how we were approaching things. There was a long-term benefit to that.

It was really beneficial to us at CHIPS. There were many times we brought a group of investors to DC, sat around a table with Gina Raimondo, and discussed what they were seeing China do. A lot of these hedge funds have eyes on the ground in China, trying to figure out who’s doing what, how many EUV machines are actually there. You can learn a lot.

Mike Schmidt: If a company is giving you a lot of happy talk, Todd being at a conference provided a useful check. We’d talk that night and he’d say, “There’s a lot of skepticism here about Company X being able to do that thing.” That might be right or wrong, but it’s a helpful data point.

Todd Fisher: Same with the customers. Once they realized we were credible people genuinely trying to do the right thing, the relationship changed. If you look at the Venn diagram of our interests versus theirs, it was almost a complete overlap — they needed more supply. We ended up with really collaborative discussions that gave us a lot of information and knowledge, allowed us to understand what was going on, and empowered us to push harder: “Why don’t you try this? Why don’t you try that?”

That’s what a separate national investment bank — or whatever you want to call it — could offer. The advantage of building those relationships and having that credibility would be really valuable. That’s something most people don’t focus on when they think about the kind of work we ended up doing.

Mike Schmidt: But I’d guess that folks at the Department of Energy working on rare earths have developed exactly that. They probably know the ecosystem well, know the targets. I bet they’re doing it all.

Jordan Schneider: An interesting contrast is that you two did not come at this with deep sector experience — you hadn’t worked in the chips industry for 30 years. You added people to the team who did have that background, but the leadership came from elsewhere. On the R&D side, it was different — there were leaders who had spent 25 or 30 years living and swimming in this world. Without speaking about them specifically, what were the advantages and disadvantages of coming at this without having spent 20 years at Intel or investing in chip stocks?

Mike Schmidt: The disadvantages were very clear. We had to learn a lot in a very short period of time and develop credibility. I had a very explicit conversation with Secretary Raimondo about this when she offered me the job, because I wanted to be fully transparent with her — I did not have a background in some of the domains that seemed relevant for the role.

What she said was, “For this role, I want someone who knows how to get something done in government. That is its own craft.” She said I’d figure out what I needed to figure out, and we would build the team. To build that team, you needed a whole bunch of things — commercial financial expertise on one hand, and industry expertise on the other. That ended up being the project.

Really importantly, building the team was a department-wide, administration-wide effort. Todd came in as Chief Investment Officer. Donna Dubinsky, a very successful Silicon Valley entrepreneur, had been recruiting people for months at the Commerce Department and handed us a list saying, “I think you should hire these people.” Many of them did end up on the team.

In the early days, we had monthly meetings in the West Wing with Brian Deese and Jake Sullivan. The first topic on the agenda was always the team. If there was a target we were trying to recruit, we would lean on them for help — “Hey Jake, can you take your phone right now and text this person? Don’t put it away until you do.” And they would.

Raimondo — one of her many talents is attracting talent. To her tremendous credit, she was obsessed with getting good people in, and once you identified someone, she was obsessed with closing them. There was never a time where I said, “Secretary, I need you to take 15 minutes with this person to help me close the deal,” and she didn’t do it. All of that came together to build a pretty awesome group of people with the expertise we needed.

Jordan Schneider: From your perspective, the upside was that you can’t be a semiconductor expert and know how to get things done in government?

Mike Schmidt: You could be, but that person didn’t exist. Maybe they did and maybe they would have done a better job.

Todd Fisher: Whether you have semiconductor experience isn’t really the point. As we discussed earlier, the deck is stacked against you. The way government is currently set up, the deck is stacked against you when it comes to getting big things done in a compressed period of time — particularly when you’re starting from scratch and have two, two and a half years to build a team, build a process, build a system, get money committed, and start getting it out the door. That’s a Herculean task even outside of government. Inside government, the odds are even worse.

Part of what Competing with Giants is all about is how to unstack that deck so these things become easier to do in government. The critical thing from a leadership perspective isn’t just knowing how to get things done in government — though that’s part of it — but having a maniacal focus on project management, prioritization, pushing things forward, breaking down barriers, operations, and process. That’s the thing. If you also happen to be a semiconductor expert, great — but you can find the semiconductor expertise separately.

I’m never going to be a pure semiconductor expert, but we learned enough to have credibility. That’s the lesson from my career in private equity. Over the years, I did retail deals, chemical deals, financial services deals. You learn what makes a good investment, what questions to ask, and how to be a curious person — who do I need to find the answer from, how do I network, who do I bring onto the team or contract as consultants to answer the questions I know I need answered but am never going to be smart enough to figure out on my own?

Mike Schmidt: That’s the trick of investing. On the question of expertise, a couple of things. My successor, a woman named Lynelle McKay, came from the industry with 25 years of executive experience. She stayed for close to nine months after I left, serving as director of the program in the Trump administration. She had that industry background plus a couple of years of government experience working on our team. She did unbelievable work and then led the team. There are different profiles that can work — but in that startup phase, having government execution capability was probably really important.

The other thing I’d say is the semiconductor expertise we needed didn’t just come from the industry. Perhaps one of our most valued experts on specific technologies came from the intelligence community — deep, specific knowledge of how different technologies would be used in different national security contexts. On every deal we did, we’d go to him to understand why a particular technology mattered.

Todd Fisher: We also had an investor on the team who had been doing semiconductor investment — he’d never been in semiconductor fab manufacturing or engineering, but he had a completely different network of people. That breadth of networks across different parts of the economy gave us a lot of advantages.

Subscribe now

On America’s Chemical Industry

Jordan Schneider: Let’s take a chemicals detour first. Chris Miller was texting me the other day: “Chemicals is an industry no one in the policy world understands. China’s crushing it, and it’s probably on the cusp of major transformation and R&D driven by AI. The traditional petrochemicals are being crushed by China. BASF is losing production. Korean players are non-existent. But America is starting to be more viable due to low natural gas prices.”

Todd Fisher: He’s talking about a very specific part of the chemical industry. Petrochemicals in East Asia involve massive capacity. But when you think about chemicals for the semiconductor industry — which is all upstream — that’s not solely, but significantly, about very small specialty chemicals like photoresist esters, for example. That’s a whole different way to think about it.

Mike Schmidt: Very, very different. We struggled a little with those because they didn’t get the investment tax credit. To the extent we could incentivize them directly, those were mostly smaller deals that didn’t get done while we were there and are still sitting on the table. The hope was that by creating enough of a demand signal through operations at the major fabs, the suppliers would want to co-locate over time.

Jordan Schneider: Todd, when we talk about doing industrial policy in giant, sophisticated industries versus the small, struggling ones — when you think about the people you had across the table at Intel, TSMC, and Micron, these are high-flying industries with very strong competitive dynamics. The execution risk profile is completely different from a mine that hasn’t existed for a while and may not even be viable. From an investment perspective — something you’re always thinking about in private equity — how did that factor into the CHIPS Act work, and how would it need to factor into the long tail of industrial policy challenges?

Todd Fisher: It factored in very significantly. Maybe the first conversation — certainly one of the very early conversations I had with Secretary Raimondo — was: “We’ve got the world’s most sophisticated people across the table from us. They have unlimited resources and can hire the most sophisticated advisors and lawyers. We need to face off against them in a highly credible way.”

That shaped how we hired. It shaped team structure — something Mike and I talked a lot about up front. There are many different ways you can structure a team, and we ultimately chose to organize it more like a typical investment firm. Small, dedicated teams facing off against each company — one team deeply getting to know Intel, another for TSMC, another for Samsung, and so on.

It also shaped the kinds of advisors we brought in — a whole other topic in government, and a crazy process. But I was very focused on making sure we had an advisor by our side, so that if a company was going to bring Goldman Sachs to tell us why we were wrong, we had someone of that caliber on our side.

A lot of the upfront design work connects to what we discussed earlier about whether legislation should give flexibility or hem you in. That flexibility allowed us to step back and say: given this industry, this set of applicants and customers, what do we need? How do we structure the team, hire the right people, and create a process that ensures our independence?

It also drove some specific design choices. I’ve written about some of these, but our upside-sharing mechanisms — a form of equity — were designed not only to protect taxpayers and capture value if companies did far better than expected, but also to align information asymmetries. We wanted to ensure there was no incentive for companies to tell us the numbers were either too high or too low — there was an incentive to give us the real numbers. We tried to structure the overall program to limit information asymmetries and to be able to sit across the table from these highly sophisticated companies as sophisticated counterparts ourselves.

Jordan Schneider: Thinking about the small-cap companies — one of the big tools private equity firms have is that when they take something over, they can bring in new management. As we start doing more of these equity deals, is there a universe where we really need a company to succeed but don’t trust the person running it, even though they’re the only one with the capital? We don’t have to do the Intel analogy right now, but what happens when you need a firm to do something and the leadership you don’t think can get you there?

Mike Schmidt: There’s a question of legal authority. Does your deal give you the authority to change management? Most of these deals would not. But then there are informal mechanisms — obviously, tweets can do things. What I would say is it should be a very, very high bar. Extremely high. I wouldn’t say never, but the government should have a very high degree of humility about whether it knows enough to decide who is best to manage a company. The strategic interest may never be high enough where you’d feel compelled to do it and confident enough in how to execute it. There are precedents — the auto bailout, where they changed management — but that’s the rare exception.

Todd Fisher: It’s the absolute exception. With CHIPS, we were on average giving these companies 10% of the cost of a project. There’s a whole separate topic of whether you should own direct equity in these companies versus some form of warrant or upside sharing that’s more taxpayer-protective, as opposed to picking winners and losers, actually owning shares, or having voting rights. You get into a very slippery slope, because I don’t think government should be in the business of business in that way.

Government should be trying to figure out what the market is failing to provide — whatever externality or other reason the private sector isn’t willing to fill — and then use government funds and tools to incentivize and shape the market.

We had so many debates about TSMC, Intel, and Samsung. Obviously, TSMC is by far the market leader today. You could argue: put all the money into TSMC because they’re going to get us there. Or you could argue: don’t give any money to TSMC because they’re a Taiwanese company — we should be supporting our own. Our conclusion was that we need TSMC to be successful. They have the know-how. They can help us build an ecosystem. But it’s not a healthy market if only one player can really serve it. We should be encouraging the other two — because there are only two, Samsung and Intel — giving them every possibility of succeeding, and then letting the market sort itself out. That’s a much healthier environment where you’re creating the ground rules and the set of incentives, enabling these companies to compete effectively, and letting the market function.

Jordan Schneider: Let’s talk about the emotional relationship to all this, now a year out. Do you drive past the TSMC fab with pride?

Todd Fisher: When you think about the transformational potential, it is mind-bending. If you go to Phoenix and see the pieces of the puzzle coming into place — TSMC, Intel, Amkor, the supply chain, ASU, the local and county governments, the innovation ecosystem developing around it — you can look ten years out and say this could be the equivalent of Hsinchu. That’s incredible.

Share

When you think about what could happen in upstate New York — Micron just broke ground a few weeks ago — the scale that comes with memory and the ongoing investment could totally transform the region. It’s going to take a decade or more, but it’s pretty exciting to think about planting those seeds. Same in Taylor, Texas. These are potentially really transformational investments.

Mike Schmidt: We felt a huge amount of pride. We got our deals done, we were catalyzing investment, and it really was on a good trajectory. But at no point in the process — until maybe a month before the end — did that seem self-evident. It was a grind. The release of saying, “We’ve done our part and we can pass the baton,” felt really, really good.

Then there was definitely a period of uncertainty about this organization and program I’d spent two and a half years obsessing over every day. For a few months, signals about the program’s future would come through — a statement in a speech, different indicators — and I didn’t have the emotional distance to treat any of that with detachment. I just felt attached.

On top of that, DOGE came in and ended up laying off maybe 40 or 50 people. Those were friends and colleagues who joined the team because they believed in the mission — awesome public servants. From a human perspective, that was really tough.

Over time, the program has stabilized. The investment tax credit expanded. Broadly speaking, there’s been significant reinvestment in the program.

Todd Fisher: That’s a little-understood development. The investment tax credit was expanded by 10% in the Big Beautiful Bill. That’s more money than we had to invest through grants.

Mike Schmidt: They also haven’t imposed any tariffs on chips. The actual strategy thus far has been to increase supply-side incentives. We’ll have to see the investment trends over time, but there has been a measure of continuity, and that’s a really good thing — to the administration’s credit, because this is hugely important for national security. Increasingly over time, you just become a citizen like everyone else, paying attention to this part of the world along with all the other parts going through turmoil.

Todd Fisher: For me it was a little different, because Mike left on January 17th and I agreed to stay until March 1st. Those six or seven weeks were challenging. I’d never been through a transition. The day before I was meant to leave, the order came down to lay off all probationary employees, which for us was almost everybody because there was a two-year probation period. Anyone from the private sector who had joined was basically still within that window.

My last 24 to 48 hours were spent fighting to save as many people as we could. We ultimately saved a good chunk — maybe a third or so. But to leave on that basis, particularly on CHIPS, where most of these people had given up careers in the semiconductor industry and the private sector — to feel like they were being vilified, accused of trying to game the system, wondering whether they’d be fired, whether they should take the fork — it was a really difficult environment. When I left, I was shell-shocked. It was a tough, tough time. Everything after that, as Mike said, has been my experience as well.

The Trump administration loves equity stakes. Only one year into this term, the federal government has utilized some form of equity instrument in multiple deals with companies across industrial sectors: MP Materials, Intel, Lithium Americas, U.S. Steel, Trilogy Metals, Vulcan Elements, and ReElement Technologies. The administration has reportedly considered additional stakes in other companies, and some officials have even floated participating in the upside of patent revenue derived from government grant funding. Not since the Great Depression has the government taken ownership stakes in private corporations at such scale and speed, and in this case, without explicit Congressional authorization.1

The political reception has been mixed. Stalwart allies have criticized President Trump for what they see as unprecedented government intervention in markets, while progressives have cheered the taxpayer benefits of government investment.

Equity investments are now a visible tool in the government’s industrial policy toolkit. This is a positive development: used judiciously, equity instruments can support industrial policy goals and protect taxpayers from asymmetric investment risks. But without a clear purpose, appropriate structure, and defined exit strategies, equity stakes risk becoming “rule by deal“ — one-off transactions that benefit particular political interests rather than the public.

This piece proposes a four-part test for evaluating government equity investments: (1) whether there is defensible legal authority; (2) whether there is a clear purpose for intervening; (3) whether another tool could better support that purpose; and (4) whether there is a predetermined exit strategy to avoid bad incentives.

Equity can take different forms, each with different advantages and risks

The Trump administration has deployed a variety of means to acquire stakes in companies:

• Direct equity capital, which gives the government an ownership stake in the company.

• Equity options, which grants warrants — options to purchase equity stakes in the future at a fixed price.

• Veto power, which grants the federal government a say over certain business decisions even without an economic ownership stake.

The federal government’s use of equity isn’t inherently good or bad, but must be evaluated on its ability to minimize the downsides and maximize the benefits.

Why equity can make sense

Direct equity capital can be patient

Lending is often the prototypical tool when the government is filling a private sector financing gap. For example, the Office of Energy Dominance Financing (previously known as the Loan Programs Office) lends or guarantees lending for companies that have demonstrated technological viability but lack existing revenue streams or established credit profiles to access traditional project financing. Since these companies are closer to the steady revenue generation needed to service interest payments within a reasonable timeframe, the government can step in to provide a loan or loan guarantee. In this case, government lending can serve as a bridge from innovation to commercialization.

But there are situations where companies need a more patient form of capital. As Daleep Singh recently noted:

“There is a class of investments in projects or companies that require a lot of upfront capital investment, a very long time to generate a commercially attractive return — let’s say 10 years-plus — and that have a lot of risk that investors feel uncomfortable modelling. It could be geopolitical or regulatory — some type of risk that the government knows more about than they do. A lot of these are deep tech or physical hardware investments. The venture capital community tends not to fund these projects at pace and scale. But these companies require equity because they don’t yet have cash flows to service debt. That is the sweet spot of where equity stakes make sense.”

Adding debt burden could worsen a firm’s balance sheet precisely when it might need flexibility to continue making upfront capital investment. Innovative companies at the technological frontier often have uncertain timelines to profitability. “Learning by doing” in hard tech manufacturing requires time that debt financing doesn’t often allow. Patient capital can make the government a strategic partner for firms whose success strengthens national or economic security. In the context of targeted industrial policy, equity investments can offer support without creating additional fixed obligations. Grants and refundable tax credits can also serve as patient capital, but they are expensive — they defray costs for recipient companies but offer no prospect of compensation.

Equity options can correct asymmetric risk and protect the taxpayer

When the government makes or guarantees a loan, the most it can earn back is the principal plus interest, but at the risk of absorbing the full cost of the loan. That risk is compounded by the fact that the federal government’s forays into lending tend to be in riskier domains where the private sector is absent or less likely to invest, such as in nascent technologies or beleaguered but critical companies that need a bailout. That puts the taxpayer on the hook for asymmetric risk — capped upside but complete downside.

This asymmetry was the justification for the federal government attaching an “equity kicker” (warrants) to its loan bailing out Chrysler in 1980. Rep. William Green (R-NY) said as much in his defense of the provision:

“The equity ‘kicker’ that Congress insisted on is entirely consistent with the high risk; there is no reason for surrendering a penny of it... when a private entity provides a service and takes an economic risk, it demands and receives financial benefits. Why should the taxpayers, who provided a vital service and took a great gamble, be denied the same right?”

Chrysler ultimately repaid its loans seven years early, and the government made approximately $300 million profit from the kicker.

The 2008 global financial crisis bailout followed a similar script — the statute itself required that warrants or other instruments:

“be designed… to provide for reasonable participation by the Secretary for the benefit of taxpayers, in equity appreciation…; and to provide additional protection for the taxpayer against losses from sale of assets.”

The principle is straightforward: if taxpayers bear considerable downside risk, they deserve adequate upside participation. Equity options, such as warrants, can solve that.

Foregoing equity capital puts us at a geopolitical disadvantage

Our reluctance to exercise equity as a tool has placed the US at a disadvantage with its strategic competitors and adversaries. Equity is not inherently superior to the tools we typically deploy, but our competitors can offer more attractive financial options to emerging countries looking to build out their energy and industrial infrastructure.

Take the case of nuclear energy. China and Russia compete via “build, own, and operate” models, providing 100% of the equity and debt to construct nuclear power plants. With lending as its only offering, the US can’t present a favorable option to already debt-saddled countries looking to expand access to reliable energy. When a deal requires patient capital or greater risk-sharing, the absence of statutory authority for equity investments impedes US competitiveness in strategic sectors.

The Trump administration and Congress moved to address this gap in 2018 by authorizing the then-new Development Finance Corporation to make equity investments. Unfortunately, as Employ America has previously written, that tool was underutilized because federal budget rules treated equity investments as grants.2

The US need not exercise every tool our competitors deploy. But categorically excluding equity from our toolkit, regardless of whether it’s the appropriate instrument, creates an artificial competitive disadvantage when equity may be a more attractive (or necessary) form of financial support.

Veto power can help the government exercise control in extraordinary circumstances

The federal government may want to exercise control over business decisions in certain extraordinary circumstances. During the Great Depression, enormous failures in governance across the nation’s banking institutions had eroded public trust in the financial sector. As the Reconstruction Finance Corporation (RFC) took enormous steps to recapitalize the banking industry, it also used its voting rights to change banking practices and reduce moral hazard. The RFC held voting rights in thousands of companies across the nation and exercised them to impose compensation limits and replace officers of corporations. In this case, control was established via the voting rights attached to shares.3

The risks of equity

Government could direct business decisions inappropriately

The most common criticism of the government taking equity is that the federal government’s incentives may be misaligned with those of the company. Equity capital in the form of common stock ownership carries voting rights, and ownership may entitle the federal government to veto power over decisions as wide-ranging as board membership, executive compensation, or investment priorities. There’s a risk that government input on these decisions would be informed by political considerations or bureaucratic preferences rather than commercial logic or shareholder value. This kind of micromanagement could undermine the very competitiveness the investment is meant to support.

A recent intervention had the government acquire a veto power without any actual economic ownership, the so-called “golden share” in U.S. Steel. This kind of veto could offer a potential middle path: the government retains veto power over critical decisions without assuming the broader responsibilities and risks of being an economic shareholder. This could be valuable when foreign ownership or incentives to expand market access to adversaries create persistent national security concerns that require ongoing oversight rather than one-time contractual commitments. But this power can be wielded inappropriately in the conduct of industrial policy.

The golden share was negotiated as part of the Committee on Foreign Investment in the United States’ (CFIUS) review of Nippon Steel’s acquisition of U.S. Steel, under the authority of Section 721 of the Defense Production Act. The statute grants the government authority to review foreign acquisitions for national security risks and to negotiate mitigation measures.

But these measures are typically structured as binding contractual agreements, such as commitments to restrict technology transfers abroad, maintain data within US borders, or submit to government screening of board appointments. The statutory language authorizes CFIUS to “negotiate, enter into or impose, and enforce any agreement or condition” to address national security risks, which can venture beyond mere contractual terms.

National security is a broad concept and accordingly opens the door for abuse. The golden share allows the US government a veto power over any attempt to waive or reduce planned investments or to relocate jobs — two restrictions that are, at best, of questionable relevance to the national security of the United States.

Equity investments could signal favoritism

Direct equity capital also risks signaling government favoritism — if the government owns a stake in a company, it may be incentivized to push government contracts or grant funding towards that company. The perception of an advantage arising from a government equity stake could also pressure companies to seek equity from the government even absent a need for such capital, simply to maintain parity in a competitive ecosystem. This bolsters the critique that industrial policy incentivizes companies to lobby for individual deals. Additionally, without systematic criteria to provide government equity decisions may be, or even appear to be, political. Equity investment increases the risks of industrial policy appearing ad hoc, with no clear principles distinguishing when equity is appropriate versus other tools.

Equity investments could create incentives to stifle competition and innovation

Seeking to preserve the value of an equity stake could also incentivize the government to make bad policy decisions to avoid weakening its position. Moreover, equity investments are typically in individual companies and can undercut the market competition that breeds innovative, successful companies. As Peter Harrell and I wrote in Bloomberg’s Odd Lots newsletter on the MP Materials deal:

“The U.S. decision to back MP sidesteps this competitive process, effectively granting a monopoly franchise in magnet production. This risks locking the U.S. into a suboptimal path if MP fails to deliver on cost or performance, while crowding out rivals that could prove more innovative. Niron Magnetics is a Minnesota-based startup that is developing rare-earth-free magnet innovation — offering the promise of eliminating a key supply chain vulnerability via innovation.”4

Dean Ball recently wrote of a similar hypothetical: the pressure to maintain the value of the equity stake could intensify if some new entrant “obviates the need for the industry as a whole.” In a capital-scarce environment, backing a single incumbent could stifle competition and innovation. The strategic question is whether scarce public and private capital is going to firms with the best technologies. An equity stake rather than a competitive process risks the US missing an opportunity to foster a genuine race to the top. This would undercut a key goal of industrial policy as recently illuminated by Todd Fisher — effective industrial policy should have “a plausible corrective path that leads to a self-sustaining outcome.”

Shareholder interests might diverge from national security priorities

Shareholder interest and national security interests may not move in lockstep. This was an acute tension in the Commerce Department’s equity stake in Intel, as Mike Schmidt and Todd Fisher wrote last year in the Wall Street Journal.

Intel’s products business, which designs chips for computers and servers, is large and profitable, with a 26% operating margin. But the CHIPS Act’s incentives program was established to support domestic manufacturing — a critical national security priority. The government was primarily interested in Intel’s foundry business, the manufacturing arm that could produce chips for other companies. That business lost more than $13 billion in 2024, more than $10 billion in 2025, has almost no external customers, and carries a -58% operating margin. Because the government took equity, taxpayer exposure is now tied to overall company valuation, which is predominantly driven by the products business, and not the foundry business that justifies the intervention. If an investment decision created conflict between the successful products business and the weaker foundry business, it would put the federal government in an unenviable position: improve national security but harm taxpayers by weakening the successful business segment, or worse, strengthen the government equity stake at the expense of our national security.

Consider the following hypothetical: if Intel’s products business develops a cutting-edge chip design in the future, the government’s shareholding interest would benefit from exports to China, despite a relatively enduring, bipartisan consensus to limit high-end chip exports to China. This conflict would place policymakers in a challenging position.

The transactional nature of equity stakes also creates new vulnerabilities. With the US government as a significant shareholder, Intel becomes an attractive target for economic coercion. China could direct state-affiliated companies to stop purchasing from Intel, weaponizing the equity relationship to harm a government-backed competitor. What was intended to strengthen a strategic asset instead creates a new pressure point for adversaries to exploit.

A four-part test for evaluating government equity stakes

Given the potential benefits and the inherent risks to government equity investing, I propose a four-part framework for policymakers to evaluate whether equity is appropriate for a given objective.

1. Is there legal authority?

A government agency should establish basic legal authority and communicate it before utilizing an equity instrument. Currently, only the Development Finance Corporation has the explicit authority to provide equity capital. Absent explicit authority, agencies can defensibly argue that certain implicit authorities allow for the use of equity instruments. For example, as Employ America has previously argued, the authority to attach equity-like instruments inherently lies with the statutory requirement to protect taxpayer interest and to support deals with a reasonable prospect of repayment.

The administration has also exercised implicit authorities for equity purchases, most recently in the MP Materials deal, invoking the Defense Production Act’s (DPA) Title III authority to make “purchases of or commitments to purchase” defense resources.5 The statute also allows the president to “make provision” for a variety of purposes — a broad (but not explicit) authorization that would allow an equity purchase. In the case of the Intel stake, Commerce could be relying on an expansive interpretation of its “additional authorities” in 15 U.S.C. 4659, which include requiring recipients to “make payments… as a condition for receiving support” or to “enter into other transactions as may be necessary.”

Regarding these executive interpretations, Peter Harrell recently noted in Lawfare that “the executive branch may well be able to act unless Congress decides to intervene…” Ultimately, to place the use of equity instruments on firmer ground, Congress should grant explicit authority to agencies.6

2. Is equity serving a defensible purpose?

The government should articulate a clear purpose or policy objective that the equity investment is intended to solve. With CHIPS, the aim was to incentivize domestic semiconductor manufacturing by building a domestic ecosystem, typically by overcoming cost differentials that make domestic production uncompetitive with Asia or elsewhere. Here, equity capital is less likely to make sense.

Only after establishing a precise goal can policymakers evaluate which tools are most effective for achieving it. The critical question is: how does equity support the overall policy mission, and what type of equity instrument best serves that purpose? The answer will depend on whether the goal is to correct asymmetric risk, provide patient capital, or exercise a veto power. Different purposes may call for different equity structures, common stock, preferred stock, warrants, or other instruments, each with distinct implications for returns, control, and alignment with policy goals. But the specific purpose must be articulated clearly in relation to the broader policy objective, and the structure of the equity investment should follow from that purpose.

The role of equity will vary depending on the policy mission. In the CHIPS context, Intel can easily raise equity in private markets; offering them $8.9 billion in direct equity would not address the core challenge of defraying the cost differential of building in the United States.

For established companies with access to capital markets, equity would more likely serve a risk-balancing function rather than a financing one; that goal is better served by warrants than by a direct equity stake. In CHIPS R&D programs, equity might serve a different purpose entirely. Nascent companies struggling to raise private capital may benefit from a government equity tranche, especially if the government is willing to take risks that typical equity investors won’t. Here, government equity could catalyze additional private investment and help promising technologies survive the valley of death.

3. Could another tool, or combination of tools, better achieve the same purpose?

Different tools serve different purposes. And the government has a vast toolkit at its disposal: tax credits, loans, demand guarantees, regulatory policy, skilled labor programs, and equity capital and options. Equity instruments can serve unique purposes that other tools typically can’t, and sometimes only a combination of all these tools can achieve the intended outcome.

But companies can also raise equity privately. And a combination of tools, such as a loan coupled with a purchase guarantee, could accelerate that process. Where practicable, it may be preferable to deploy other tools (such as loans, purchase commitments, or tax credits) as necessary and avoid the government taking an ownership stake.

In the case of MP Materials, the amount and variety of support the company is receiving is staggering: a price floor for neodymium-praseodymium oxide (NdPr), a guaranteed offtake for finished rare earth magnets, guaranteed EBITDA, a loan, all in addition to the equity investment. Each of these is serving a unique purpose in the value chain, but with this level of support the company could potentially raise capital privately given the level of government guarantee. Of course, this would put the government in the position of socializing losses associated with some of the tools (such as the loan) while subsidizing the investor’s gains. That’s where the use of an equity instrument may be appropriate, and the CHIPS office included “upside sharing” without taking an equity stake in the company.

Given the complexity of modern finance, the construction of these deals is more art than science. But the threshold for using equity when another tool or combination of tools could serve the same purpose should be high given the inherent risks.

4. Is there an exit strategy to minimize political or cross-policy pressure?

Designing an exit mechanism that operates independently of political interference reduces pressure to retain an interest past the point at which the objective has been achieved. If the sale process is somewhat automated and insulated from political control, cross-policy pressure would be minimized because there would be no “decision” to make.

The exact exit mechanism should be tailored to the type of investment and aligned with the purpose of entry. For venture-like investments at the technological frontier where taxpayers deserve upside for bearing early-stage risk, the government’s shares could be structured as the first to be sold in an initial public offering. This creates a natural, market-driven exit without requiring ongoing discretionary decisions about timing or price. Alternatively, if the capital was intended primarily to be “patient capital” for some technological breakthrough on the road to commercialization, the trigger might be achieving commercial viability by some predefined production metric or when the technology is successfully deployed at scale. The key is establishing these metrics upfront before an investment is made, and automating the exit process as much as possible.

For investments where the government holds a large stake and a single sale could have significant market implications or force a discounted transaction, the exit could involve a timely disposition contracted to a private entity with expertise in managing such sales. The Troubled Asset Relief Program (TARP) provides a useful model — there, Treasury retained decision making authority but contracted with private asset managers to systematically dispose of equity stakes acquired during the financial crisis, removing day-to-day political considerations from sale timing. In another example, Traxys, the metals trading company, was contracted to bring over 7,000 metric tonnes of UF6 uranium to the market following the decommissioning of the Portsmouth, New Hampshire nuclear facility, and in a way that minimized market disruptions.

Predetermined exit strategies become far more complicated when things go wrong, and in a portfolio of early-stage or high-risk investments, at least some are bound to fail. What happens when a company misses predetermined milestones but has made genuine progress? Does the government exit mechanically, potentially destroying value and undermining national security goals, or does it exercise discretion, opening the door to the very political interference the exit strategy was meant to avoid? What about follow-on investments? If a portfolio company needs additional capital to survive, does the government participate to protect its initial stake, potentially throwing good money after bad? Or does it allow dilution or even failure, risking the loss of strategic capabilities? These decisions are inherently discretionary and politically fraught, and predetermined exit strategies alone cannot solve them. The government will inevitably face judgment calls that pit financial interests against strategic objectives, and no amount of upfront rule-setting can eliminate that tension entirely. That is all the more reason to ground the decision-making in a structure immune from other political pressures.

Strategic government investment is different from sovereign wealth fund management. Strategic investments often require concentration, big bets on specific technologies or companies deemed critical to national security. But over time, as objectives are achieved, concentration should give way to exit. The government should not be in the business of long-term portfolio management of individual companies. Once the strategic purpose is served, whether that’s correcting asymmetric risk, bridging a capital gap, or establishing a viable domestic industry, unwinding should be maximally automatic and depoliticized.

Conclusion

Government equity investments need not be viewed as an ideological transgression. When structured thoughtfully, with clear legal authority, defensible purpose, consideration of alternative tools, and predetermined exit strategies, equity can advance national interests, strengthen critical industries, and protect taxpayers from asymmetric risk. The challenge is not whether to use equity, but how to use it well.

The Trump administration’s embrace of equity instruments represents a significant shift in federal economic policy, one that requires careful guardrails to prevent “rule by deal” dynamics. But categorical opposition to government ownership overlooks equity’s unique capabilities in certain circumstances: providing patient capital where debt is unavailable or would unduly constrain capital investment, correcting risk asymmetries where taxpayers bear downside exposure, and maintaining oversight in extraordinary situations. With appropriate frameworks that minimize political interference and maximize strategic value, equity can serve as an effective tool in the government’s industrial policy arsenal without compromising market dynamism.

But the thrill was short-lived. Preliminary term sheets are not final awards. To fully implement the program, we still needed to sign binding contracts with these leading-edge companies and close another dozen or so deals. The contracts would commit government subsidies (usually around 10% of total capital expenditures) to support specified projects and detail the milestones and other conditions for receiving those funds over time. And that meant, among other tasks, defining the legal relationship between the government and the applicant, which would govern the funding process and establish the legal rights of each party.1

I didn’t anticipate that establishing this legal relationship would become the central preoccupation of the rest of my tenure as program director. I expected these contract terms to be negotiated primarily between lawyers — complicated conversations, no doubt, but ultimately the kind of thing lawyers work through. Instead, our proposed terms drew fierce opposition that escalated well beyond the legal teams. Company executives and boards became directly involved, viewing these provisions as fundamental business issues rather than technical legal matters. It became clear that, absent concerted negotiations that included executive leadership, we wouldn’t close our deals, and the program would fail.

Over the coming months, Todd and I immersed ourselves in the legal details alongside a group of truly exceptional Commerce Department lawyers. These details were deep in weeds familiar to transactional lawyers, allocating risk and protections across a range of possible adverse outcomes. And the negotiations were intense — a negotiating session with a single company might stretch across a full day or even a full week, and several sessions were needed to close each deal.

We ultimately succeeded, closing 20 awards totaling roughly $34 billion in direct funding and $5 billion in loans by January 2025. It was hard-earned, but with the benefit of hindsight, we wish we had thought more holistically about these issues as both legal and policy matters earlier in program design. This piece describes some of the key challenges we faced to help future policymakers anticipate and navigate them. In the process, it may illuminate a new dimension of what industrial policy looks like at the ground level.

What informed our legal framework?

The challenge for the Department’s legal team was that there was no clear precedent for the types of discretionary, grant-like industrial subsidies we were providing. So the team drew on other sources.

The most influential precedent was the form loan agreement used by the Department of Energy’s Loans Program Office. LPO’s loans are a direct support for industrial projects and seemed like the strongest analog to what we were doing, but they were an imperfect one: loans must be repaid, but our support generally would not be (with exceptions, as discussed below).

The team also drew from grant regulations, but these too were an imperfect fit. In the US government, grants typically go to state and local governments, or nonprofits like universities; when they do support industry, it’s usually at a relatively small scale. Our agreements also had to integrate a range of statutory requirements that applied to CHIPS funds specifically or federal financial assistance generally. And, of course, we had to account for our bespoke program objectives.

Our legal team compiled this framework into a form award agreement that ran around 150 pages, which would evolve substantially over the subsequent months of negotiations between the government and industry, between the CHIPS office and counsel’s office, and between counsel’s office and the DOJ.2

The following eight challenges proved among the most difficult to resolve.

Subscribe now

Challenge #1: What are the consequences of violating laws unrelated to the CHIPS program?

At the most basic level, our award contracts were commitments by the government to fund projects in tranches over time. The money wouldn’t go out up front, but as construction, production, technical, and/or commercial milestones were met. But beyond hitting those milestones, we needed to decide what standards of conduct, if any, the companies needed to maintain to be entitled to ongoing funding. A baseline question related to their obligation to comply with laws unrelated to the CHIPS program. Like other government financial assistance and loan programs, our agreements had a general compliance-with-law standard. They also included specific provisions related to compliance with laws governing things like sanctions, export controls, intellectual property rights, labor law, and environmental law.

These provisions track a pretty strong normative intuition: that a government probably shouldn’t be forced to deliver on potentially billions of dollars in funding if a company is breaking the law. We had to ask ourselves, for example, whether the government should be able to withhold funds if an environmental catastrophe or sanctions violation comes to light. But for companies, these provisions raised the prospect that minor foot faults could become excuses to withhold the funds needed to make the project economics work. And for projects of this scale — tens of billions in investment, thousands of employees, dozens or hundreds of external suppliers — such foot faults are common. Applicants also pointed out that the US government has existing enforcement tools for all these issues, and that CHIPS incentives had a different purpose — to encourage chip manufacturing. These arguments resonated with many on our team and informed our negotiation posture.

We ultimately settled on a middle path: holding the line that the companies must comply with law to maintain compliance with the agreement, but negotiating the specific language to ease their concerns. We didn’t take the extreme position that compliance with other laws should be left entirely to other enforcement regimes, but we also wanted to make clear that foot faults wouldn’t derail investments. We wanted to “keep the main thing the main thing,” and in this case the main thing was encouraging semiconductor manufacturing.

To put this into effect, we qualified the legal requirements with materiality standards to codify that our intent was not to penalize minor infractions but to give the government the latitude to withhold funds for major ones, even if milestones were met.3 We often settled on a simple fix: adding the word “material” to a provision to raise the threshold for penalty. For example, only “material” violations of environmental or labor law would trigger noncompliance. For provisions relating to maintenance of intellectual property, we clarified in the contract language that the losses of IP that would trigger the provision should only be those that would make the project no longer viable, such as a third party suing for IP infringement and winning an injunction against using IP that was essential to project viability.

Challenge #2: How do you navigate other legal requirements that attach specifically to the funding?

Solving the first challenge required defining how to treat violations of laws that govern applicants’ economic activities, regardless of whether they took CHIPS funding. But by receiving CHIPS awards, applicants were also signing up for a slew of additional statutory requirements, all of which would have to be implemented through the terms of our contract.

Congress decides how these requirements apply through either explicit stipulation or omission. In some instances, there was affirmative language stipulating requirements. For example, the CHIPS Act explicitly established the so-called “national security guardrails.” These provisions required the clawback of funds if companies “materially expanded” manufacturing capacity in China or engaged in joint research or technology licensing with Chinese entities in domains that raise national security concerns. Similarly, the CHIPS Act also specified that recipients of CHIPS funds would need to pay “prevailing wage” for construction work under the terms of the Davis-Bacon Act.

Our terms also had to account for baseline requirements that broadly attach to federal funding or other government action unless an exemption is provided. For us, this was the case for NEPA (until Congress provided an exemption) and statutes like the “Fly America Act” (which requires anyone taking a flight paid for with CHIPS monies to fly on only “US flag” air carriers) and the Trafficking Victims Protections Act (which addresses compliance with human trafficking laws). Our legal team compiled a detailed spreadsheet tracking these laws to ensure that we were addressing them as required.

Importantly for program implementers, these requirements are not self-actualizing. At a minimum, they need to be incorporated into award terms. But implementation often calls for much more than that. For example, we defined the parameters of the national security guardrails through a 68-page notice-and-comment rulemaking and then conducted company-by-company diligence for each applicant to document a baseline of existing activities in China and delineate which activities would be allowed going forward. I’ve also written previously about the challenges associated with NEPA.

We succeeded in reconciling these many intersecting requirements when we started early and engaged with industry. For example, choosing to go through notice-and-comment on the technical details of implementing the national security guardrails helped us achieve Congress’s policy goals without derailing investment in ways we didn’t anticipate. By contrast, we were caught off guard by some of the challenges of implementing Davis-Bacon, and that came with a cost.

But even with the best planning, lawmakers should appreciate that attaching additional legal requirements to government funding can have real impacts on program outcomes. They can impose administrative costs, such as systems investments for both companies and the government, and added fiscal costs because companies request higher incentives to account for additional costs or uncertainty. Stringent requirements can also deter participation altogether. We had companies withdraw applications and decline to apply because they didn’t want to comply with the statutory constraints on their activities in China.

Challenge #3: How do we address the ghost of “federal interest”?

While most legal requirements on CHIPS funds were imposed by Congress, one came from the courts.

“Federal interest” is considered a common law right, meaning that it was created by federal courts rather than statute — in this case, through a series of judicial rulings over the past century. The basic premise of federal interest is that if the government gives a grant for a project (say, a grant to a nonprofit to build a workforce center) and either (a) the grant isn’t used as intended (the workforce center becomes a commercial warehouse), or (b) the entity that received the grant goes bankrupt, the government retains a financial interest in the property that allows it to recoup all or a portion of the grant. In the case of bankruptcy, the government has “super senior” interest — skipping ahead of any other creditors. And while our awards were not technically grants (they were other transactions), case law established that federal interest attached to our form of financial assistance.4

Addressing the federal interest with our applicants was challenging. For them, the concept was foreign and unexpected: they understood our grant-like awards as not requiring repayment unless a specific clawback was invoked, so it was hard to accept that CHIPS funds would give the government a financial interest in their property. Moreover, it would constrain their flexibility to enter into financing arrangements in the future — if the government had a super senior interest in a fab in bankruptcy, financing would be harder to secure or more expensive. And for companies with existing financing arrangements, intercreditor issues would arise. How would the federal interest be treated compared to first lien lenders and equipment financiers? Settling these matters often requires executing intercreditor agreements that establish the rights of existing lenders relative to the government, demanding negotiation and agreement with banks or other third parties. These are complex questions for any award, but even more so for a semiconductor project with thousands of pieces of equipment for which the government is providing a small minority of the funding.

The application of federal interest in an industrial policy context presents policy tradeoffs. On the one hand, it does create real financial protection for taxpayers: if a company goes bankrupt, a security interest in the underlying property allows the government to recoup funds and avoid the money-losing scenarios that activate oversight (see the Solyndra hysteria). And in at least one or two instances involving companies or projects with more precarious financials, we were glad to have that taxpayer protection, which provided security beyond a typical bankruptcy-activated clawback. But in most cases, federal interest created a considerable hurdle without commensurate benefit — the probability that its protections would ever be necessary was so remote that it felt like wrestling a bear to protect a house that was miles away. Congress should consider waiving federal interest for future industrial policy programs, or explicitly empowering the executive branch with the ability to do so.

Challenge #4: When should we use the bazookas of disbursement stops and clawbacks?

Beyond determining the substantive standards we’d hold awardees to, we also needed to decide on consequences for violation. There are really only two tools that give the government meaningful leverage after an award is issued: clawing back funds already paid, and stopping disbursement of future milestone-based payments.

But applicants viewed both of these interventions as bazookas. Even disbursement stops, which are considered the less extreme option, could threaten the financial viability of a project. Applicants pushed us to limit the conditions for disbursement stops to a subset of contract violations, but we felt that this would make the excluded terms more or less unenforceable. We therefore maintained our right to disbursement stops as a general remedy for contract violations. But to address applicants’ concerns, we negotiated the materiality qualifiers discussed above, and also defined cure periods to give companies the right to fix a problem first.

The bigger challenge for us was figuring out where we’d reserve the right to claw back and where withholding funds would be sufficient. In traditional grants, clawbacks aren’t available to the government (with the exception of the federal interest, as explained above). But CHIPS mandated two conditions for clawbacks in statute: violation of the national security guardrails, and receiving funds for a project and then failing to commence or complete it. Beyond the statutory requirements, we initially included clawbacks in the set of tools available to the government, while providing verbal assurances that our intent was to only use them in extreme cases. But verbal assurances weren’t good enough for applicants — the consequences of a clawback were too dire, and the posture of the government could change with an election. We responded to these concerns and ended up limiting the clawback conditions to a narrow set of circumstances, such as abandoning the project entirely or bankruptcy.

Again, there were trade-offs here. Limiting clawbacks to a narrower set of violations limited the government’s ability to enforce the other terms of the agreement once funds were distributed (typically over the course of 3-5 years). But we understood that the prospect of clawbacks for relatively minor breaches would be commercially untenable for our applicants. We shaped our approach accordingly.

Subscribe now

Challenge #5: How bounded is the government’s discretion?

In typical commercial arrangements, parties to a contract generally have a duty to interpret and perform contract terms reasonably. In other words, if one party’s actions under a contract are challenged, it may lose if the court finds it is not meeting a reasonableness standard. This duty to act reasonably can be defined in contract terms, but even without explicit language, courts will impose it as an implicit duty under common law.5

Consistent with other federal grant agreements, our terms provided that CHIPS awards would be administered at “the discretion of the government.” But that language — which implied broad latitude — made some of our applicants very uneasy. In the event of a dispute under the contract, they wanted courts to hold the government to a “reasonableness” standard typical for commercial agreements.

But we couldn’t accommodate that request. The challenge was that the Administrative Procedure Act gives courts the ability to invalidate certain actions by government agencies (including withholding grants) if the court determines that they are “arbitrary and capricious.”6 Our legal team analyzed the APA and determined that it applied to our awards and that we didn’t have the authority to bind the government to a more constrained standard of review beyond the “arbitrary and capricious” test of the APA.

The standard makes sense in many traditional government contexts where agencies make decisions based on policy judgment. But industrial policy programs blur the line between typical government administration and commercial partnership. When the government is acting as a commercial counterparty — negotiating terms, making milestone payments, and enforcing performance requirements similar to a private investor — companies expect the legal protections typical of commercial relationships. Under the “arbitrary and capricious” standard, even if the government made a questionable interpretation of the agreement that a court might consider unreasonable in a commercial context, companies would have little recourse as long as that interpretation was within the bounds of the APA. This concern was amplified by the political reality that “the government” might mean entirely different decision-makers after an election. Companies worried they could be locked into long-term agreements with billions of dollars at stake, yet have limited legal recourse against adverse interpretations by future administrations.

For future programs, Congress should consider specifying that agencies may commit to a more constrained standard of review than the APA provides in their commercial agreements.

Challenge #6: What happens when a company wants to terminate the deal?

Many applicants also wanted the right to terminate and withdraw from the agreement. In traditional grantmaking, applicants usually seek out the grant because they have limited other sources of funding. But CHIPS grants were typically not about providing cash to companies that didn’t have it — they were about creating internationally competitive project-level financial returns for applicants that already had access to ample financing. So applicants raised the idea of repaying the subsidy and relieving themselves of the obligations of our legal agreements. They particularly wanted to rid themselves of the statutory restrictions on activity in China.

This wasn’t something the applicants considered lightly, nor did they think it likely that they’d ever want to repay their grants — after all, doing so would be expensive. They were interested in termination rights mostly as a safeguard against worst-case scenarios. Some envisioned situations in which the government refused to distribute funds to which companies were legally entitled under the terms of the award contract. Absent termination rights, their only recourse would be to sue the government, which they would be loath to do given the government’s broader power over their business. They could end up receiving none of the benefits of CHIPS funding but all of the associated constraints.

We never thought of CHIPS grants as creating a compulsory obligation: if the economics of a project no longer panned out, we didn’t want to compel completion anyway. But each award also came with a real opportunity cost — we were allocating funds for future milestone-based payments that could otherwise be directed at other national security priorities. And in that context, allowing for low-friction termination didn’t feel right. We suspected that Congress did not intend for companies to have the benefit of subsidized financing only to turn around and relieve themselves of the obligations that came with the money. And the statute constrained termination rights, especially when it came to the national security guardrails.

Challenge #7: What happens if the government gets sued?

Indemnification may seem like an eyes-glaze-over legal detail, but for us it was a matter of intense business interest. The core issue was the prospect of third-party litigation against the government: if the government was sued for the actions of a company it had funded, would the company indemnify (i.e., reimburse) the government for any liability incurred? For the government, indemnity was standard fare and a sensible protection. Why should the government be on the hook for providing funding? And in the context of loans, lenders (including government lenders) are always fully indemnified. Given the size of the funding and complexity of the projects, our legal team felt that the CHIPS program should be indemnified too.

Companies disagreed. CHIPS was typically providing only 10% of project funds, and the government gets sued for all sorts of things; could they really leave their shareholders on the hook for major financial risk years in the future? Even if the prospect of significant liability against the government seemed unlikely, the potential dollars could be large — the types of hard-to-underwrite financial exposures that companies are supposed to protect their shareholders from.

For several deals (including all of our largest and most important ones), the indemnity provision was one of the last issues we resolved, hanging over the entire negotiation and often requiring sign-off from CEOs and boards. What made these negotiations particularly challenging was that they involved a zero-sum allocation of risk that was hard to foresee. We ended up negotiating the specifics of the indemnity in a way that partially addressed company concerns, but companies still had to get comfortable with the resulting provision.

Subscribe now

Challenge #8: What happens if the company sells the project — or itself?

There was a suite of issues related to change of control, requiring us to anticipate the myriad situations in which a project’s or company’s control might transfer to another entity. Possibilities included a recipient merging with another company, selling the project to another company, or even selling a minority interest in itself or the project.

For companies considering M&A activity, having to go to the US government for “approval” to execute a transaction in their shareholders’ interest generated intense scrutiny, including at the CEO and board level.7 For example, in the summer of 2024, Intel released an earnings result that caused its stock to drop by roughly 40% and triggered intense M&A interest, both publicly and privately. Would Intel be merged with Qualcomm? Or Broadcom? Or GlobalFoundries? Would it take new outside investment in its foundry business? Or enter into a joint venture with TSMC? These questions were now of substantial import to both Intel’s shareholders and US national security: as the only US-based leading-edge logic manufacturer, Intel’s foundry business was critical to the leading-edge ecosystem we were trying to build. Intel was also working with the Department of Defense on certain national security-related programs. As a result, the change-of-control provision with Intel was heavily negotiated. Ultimately, Intel viewed the provision as so material to shareholders that it described it in detail in an SEC disclosure after we signed the deal.

The interest in change of control wasn’t limited to large companies — sales of projects and companies are common throughout the semiconductor industry. And even for companies that did not anticipate M&A activity, requiring government approval for future transactions could limit their optionality, drag on their stock, and create financial penalties.

We negotiated these provisions extensively, detailing what would constitute a change of control, whether any transactions would be exempted, what the approval process would look like, and what the consequences would be for proceeding without approval. Along with indemnity, this was often one the last terms we finalized.

Some companies understood our position but sought targeted flexibility. Others argued that we were overstepping. They felt that our agreements protected the government’s objectives through other provisions and that ownership issues should be left to companies and their boards. But we had vetted our applicants thoroughly and did not want to be bound to do business with future participants that we could not foresee or approve, and who might pursue strategies that run counter to the government’s interests. We generally held the line on change-of-control provisions while negotiating nuances and exceptions to address the specific concerns of companies without compromising program objectives.

Closing thoughts

At one point late in our TSMC negotiation, I asked one of their lawyers how our terms compared with Japan’s approach. The lawyer looked confused: “Japan? We don’t have a contract with Japan. We just submit our receipts and they give us the money.”

I was thinking about that conversation while reading Dan Wang’s recent book Breakneck. In comparing the US and China, Dan draws a distinction between what he calls the “engineering state” and the “lawyerly society.” In his telling, China’s government is run by engineers, who bring a technocratic, build-first mentality to problems both physical and social. America’s system, by contrast, is dominated by lawyers (like me!), who he says bring a more process-oriented, risk-averse perspective that tends to block progress as opposed to advancing it.

We are never going to be Japan or China — I think our political system demands protections for taxpayers in ways theirs may not, and those expectations manifest through legal frameworks. But I think Dan’s framework does have some purchase on the CHIPS experience. The legal provisions we negotiated didn’t emerge from a vacuum. They came from Congress, from courts, from regulations, from other precedents in the executive branch, and from sensible, good-willed efforts to protect taxpayers. And fundamentally, these were negotiations about risk allocation. Each provision we debated represented a choice about who bears what risk. What happens if a company breaks the law? What happens in bankruptcy? What if a company sells itself? How much discretion should government retain? These are legitimate questions with real consequences. The challenge is that each protection, however sensible in isolation, adds friction in aggregate.

The Commerce Department played its game of industrial policy Minesweeper successfully. Twenty awards, $34 billion in direct funding, $5 billion in loans — all negotiated under intense time pressure. We proved that the American system can move with commercial speed and flexibility when necessary. But the effort required was extraordinary, and I’m not sure the approach is sustainable going forward. Each legal protection created costs — in deal complexity, in negotiating time, in relationship friction, in dollars — that accumulate even when individual provisions make sense.

Future programs will face similar tradeoffs, and those tradeoffs deserve serious consideration. After all, CHIPS tax credits — roughly double our direct funding — operated much closer to Japan’s model. The question for both policymakers and implementers is finding the right balance between legal protection and execution efficiency. That’s a conversation worth having as industrial policy becomes a fixture of American economic statecraft.

I was wrong. At CPO, consultants proved essential to our success, and I’m now convinced that it would be foolish to attempt to stand up any large-scale effort like ours without them. But I had to learn how to use that outside support strategically and effectively.

Four characteristics of consulting engagements

At the outset, we brought in a mix of consultants to help us get off the ground, engaging financial and industry experts, project management support, general consultants, and a large systems contractor to stand up the application infrastructure. As the program evolved, we layered in additional specialized support: environmental consultants came on as the application pipeline materialized; an investment firm supported loan underwriting; and we later engaged a Federally Funded Research and Development Center (FFRDC), federal credit experts, and law firms. At one point, we had over 100 consultants badged and working alongside federal staff.

In reflecting on what worked well, four characteristics and decisions stand out:

1. Scope: Were the contractors bringing specialized expertise or more general support to the table? Was the work itself specific or broad?

2. Managing uncertainty: How did we account for incomplete information when defining scope?

3. Mode: Were the contractors embedded within the team, or did they lob deliverables over the fence?

4. Duration and eventual insourcing: Were these pinch hitters, or did we anticipate long-running engagements?

Scope

One clear dividing line in effective contracting was whether we brought firms in for specialized or general-purpose work.

• Specialized work, such as financial modeling, commercial underwriting, industry expertise, transaction counsel, and environmental review, was well suited to outsourcing. These were highly technical areas that required expertise that is scarce inside government, and the cost of getting things wrong is high. Contracting gave us access to talent and institutional capability we could not otherwise assemble, and materially reduced risk. The narrow focus allowed us to inject support as needed.

• General work that sat closer to the core of the program itself, such as program and project management and strategic planning, proved to be a poor fit for consultants. One case in point: early on, we brought on a large consultancy to stand up a program management function, which was to provide a view into our project plans and progress against them. In practice, the consultants lacked the requisite context and ended up building project trackers that were just not a fit for us — the trackers created the appearance of control but were very clearly false precision. We were moving quickly, and strategic ownership, prioritization, and sequencing had to live inside the program.

Managing uncertainty

When building a new program, you don’t necessarily know what you’ll need down the road, which complicates defining expectations for an engagement.

There are two common ways to handle this. One is to plan for uncertainty in throughput, bringing in support for clearly defined tasks without predefining the volume or timing. In our case, evaluating applications and supporting deal teams fell into this category. The work itself was specific but the volume was unpredictable, and building in capacity to absorb spikes and hedge against attrition proved invaluable.

The alternative approach is to enlist broad “program management” support. These are consultants who are brought in to help with unstructured problem-solving or process design. It can be enormously valuable early on, but only under very specific conditions. Effectively outsourcing program management hinges on finding exceptional operators with the right mindset and orientation, deeply embedding them within teams, and requiring clear communication and management. Absent those conditions, generalized support can become a burden to manage that costs more than it’s worth.

Mode

Consultants can be embedded and operate like part of the team, or they can work at a distance, producing specific deliverables and managed through periodic check-ins. Both modes can work, but we found that embedding dramatically increased efficacy — especially for work that was moving quickly. Embedding was essential for general unstructured work.

This was one of the problems we ran into with the program management function. Outsourcing such general functions requires shared context and foresight that the consultants could not have. Without it, an engagement is too disconnected from the pulse of the program and doomed to fail.

Embedding changes the risk profile. It won’t solve a poorly scoped engagement or make up for bad staffing, but when consultants were on site and treated like staff — included in team meetings, receiving real-time information as priorities shifted, and operating under clear day-to-day direction — we were much better able to make general support and unstructured problem solving useful.

But mere embedding doesn’t guarantee success. Effective embedding requires strong contractor-side leadership: someone who could absorb context, manage their own team, and adapt quickly without constant government oversight. When done well, the context gained through embedding enables flexibility, letting you move staff around as needed.

Duration and eventual insourcing

Over the course of the program, we had to decide whether to bring certain functions in-house. No complex organization, public or private, insources everything. It’s tactically important to distinguish which services should be temporarily outsourced and eventually folded in, and which should be enduring contractor engagements.

It’s hard to make that call at the outset. For example, outside industry experts were essential early on in orienting ourselves to the semiconductor ecosystem, helping us understand firm behavior and check our assumptions. But as we hired staff and built our own networks, that need diminished. What began as essential context-setting support became something the institution could — and should — do for itself.

Process design followed a similar arc. Early in the program, having outside support to translate ambiguity into workable processes was enormously helpful. But as we did more reps, it became clear that outsourcing the capability indefinitely would be unsustainable. We decided to learn and adapt their underlying frameworks and own the work ourselves. Unlike the gradually diminishing need for industry experts, this was a deliberate decision and required a careful transition as we brought the function in-house.

Leaders need to be intentional, early and explicitly, about which capabilities they expect to remain external and which they ultimately want the organization to own. Deferring the question makes transitions far more difficult.

Subscribe now

Procurement rules make all of this harder

Designing a fruitful contracting engagement is further complicated by the rules governing federal acquisitions, known broadly as the Federal Acquisition Regulation (FAR). Those rules exist to prevent corruption; the system is designed to limit discretion and contracting officers face serious consequences for getting it wrong. The cumulative effect, however, is a layered system that makes it difficult to get exactly what you need, when you need it, or to optimize for outcomes rather than lowest price. Even when we knew exactly what specialized expertise we needed, procurement timelines made it hard to enlist it at pace.

To avoid those delays, agencies often rely on simplified acquisition procedures or other faster pathways. But those alternative routes are still slow and come with their own tradeoffs, limiting who you can contract with and contract size. For program leaders already operating under pressure and uncertainty, these constraints compound the challenge.

I’m not a procurement expert, but I’ve seen and done enough to know that the system is a bear. Making it work requires creative procurement professionals who understand the process deeply and are willing to optimize for program outcomes, not just compliance.

And even with competent procurement staff, supporting the process requires many cycles of administrative work by program staff. Both our internal procurement function and the staff at NIST helped us find the most efficient pathways, but it still ate up a lot of program leadership’s time.

Fresh eyes help

This is purely anecdotal, but the most effective consultants were generally those with less experience contracting with the government. Teams that were frequent flyers in government work were less adaptable; it seemed to me that they had absorbed some of the worst intuitions about how government works. By contrast, firms with stronger commercial orientation and less government muscle memory tended to move faster and pivot more easily.

This may be a byproduct of the broader consulting industrial complex. Firms that have adapted to government procurement are often better at winning contracts, but that does not imply that they can operate with necessary urgency and adaptability once inside a fast-moving program; the investments they’ve made into navigating the FAR to win work have exactly nothing to do with their ability to actually deliver. I’m left with the belief that bringing in newer or less traditional players for critical work is worth the effort, even if they are harder to get through the procurement gauntlet.

A case study on effective contracting

Though I derived some of these insights as we built the program, in hindsight, it’s clear that one of our engagements embodied the principles of good contracting.

This particular contractor started working with the Department of Commerce before CHIPS was even passed; over the course of several contracts, they gained a wide remit, working on both general and highly specialized functions. They built out some of the economic models and supported development of the Internal Rate of Return (IRR) framework used in our investment process. They also provided staff augmentation to the deal teams, working alongside federal employees to evaluate applications. They were up to the task of unstructured problem solving and, in the absence of precedent, their process design expertise helped us build a complex program.

Other contractors had similarly broad mandates but were less effective. The success of this particular team came down to a few key factors:

• Most of their team was embedded with us, in the office, most days of the week. They joined staff meetings and were treated like regular feds. Their proximity enabled greater flexibility, letting us move them between process and deal support functions as needed.

• Their team was overseen by an exceptionally talented manager from the firm. We could rely on her as our main interface and rest assured that her team would be kept up to date, and that she’d prioritize work and shift resources as needed.

• The firm sent us high-quality staff across the board. I’m told that when a project is going well, it creates something of a virtuous cycle inside a consulting firm, enticing the best people to join the team. The actionable takeaway here is to do what you can to get talent you’re satisfied with — pay more for top-tier people if you need to, and don’t hesitate to demand a switch if they fall down on the job or if the person leading their team isn’t a match for your program and culture.

Even with a successful engagement, the rules get in the way and make it a real chore to keep resources that are working well. Given the success of this specific team, we absolutely needed to keep them in the mix. They were not a small business and they didn’t do much business with the government — a good example of the fresh eyes I came to appreciate. They were expensive, but they moved at pace and were highly adaptable. And so we did the dance that everyone does, bending over backwards to find a vehicle or a prime that would let us extend the services without breaking the rules.

Each time we had to re-up or extend their contract, we were forced to re-evaluate whether (and how) to build internal capacity. Despite our abundant resourcing, it always felt we had just enough to keep moving forward. We didn’t think we could afford to let these valuable players go — they were deeply integrated into the team’s work and understood the investment process, the context, the personalities, and the pace. We had growing concerns about cost and knew that this arrangement could not continue indefinitely, but their efficacy made them even harder to unwind.

Though we valued their contributions, it became clear that their process design work, while exceptional, was not unlearnable. We had done enough reps with them that I started to suspect we could carry this work forward ourselves. I didn’t want our program to only be viable if we had outside support.

This was contentious. The consultants made things easier, more polished, and faster, and rolling them off would create more work for senior staff who were already stretched. There was also a real fear that things would break — we still had a significant pipeline of deals to manage and had not yet finished designing our award phase process or the post‑award apparatus. This was the one time Todd Fisher and I got into an actual argument. This debate was less about whether the consultants were valuable and more about risk: Todd viewed the staff augmentation as insurance against missing deal deadlines in a moment of extreme capacity strain. I was weighing that risk against the growing cost, oversight exposure, and long-term dependency. We arrived at a compromise and decided to internalize the process design component but retain surge capacity for the deal team.

Without their help, things moved more slowly. Our analysis was less pristine, and we no longer had professional note-taking in every meeting or folks who could turn detailed decks overnight. But the disasters we feared did not come to pass — we did not miss critical issues that we worried only the consultants would have caught. The system did not fall apart.

Looking back, I’m glad we made the call to bring things in-house. We absolutely needed this team to start, and we benefitted from watching them lead design for many of our processes. We used our final weeks with them to closely observe what they’d done, which positioned us to take over and design later phases of our process without incident. Teams could stand to be more willing to transition ownership of key functions, and I now know to keep an eye out for functions to reabsorb.

CPO could not have succeeded without consultants. Some were indispensable. Others were underwhelming. The process sucked nearly always, even with the best procurement folks helping us out. Hindsight afforded us a few clues as to how we might do this better, and taught me quite a lot about the value of thoughtfully outsourcing to meet the moment.

Eating Policy

In business, culture eats strategy. In government, culture eats policy. Here we'll talk about the problems of state capacity (government's ability to achieve its policy goals) and how to fix them. From the author of Recoding America.

By Jennifer Pahlka

Many fear that, left to their own devices, many government agencies would likely expand needless forms, reporting requirements, and surveys. The Paperwork Reduction Act (PRA) attempts to throttle this reflex by subjecting any request for information from the public to lengthy review. The aim is to protect citizens, businesses, grantees, and nonprofits from having to provide information the government doesn’t truly need. It’s all well-intentioned, but the resulting choreography is a nightmare.1

In attempting to reduce burden and paperwork, the PRA paradoxically creates more of it within government. Compliance requires agencies to comprehensively detail and justify every bit of information they plan to collect from the public. Because of the long procedural timeline, agencies are often forced to submit this documentation well before critical program and implementation details are ironed out. The process obstructs iteration and comes with an opportunity cost, wasting effort that could have been directed at meaningful program development.

I’m not alone in wanting to sidestep PRA — recent years have seen increased legislative attempts at establishing exemptions.

During my time at the CHIPS Program Office (CPO), we needed to clear the PRA process 15 times to get approval for 10 different information collections (the official term for any type of form). I estimate that this forced our team to create about 60 new documents — some short, and some in excess of 20 pages. While we met our mission, who knows what more we could have done with that time?

I could live with a tedious process that achieves its intended outcomes, but it seems clear that that’s not the case with the PRA. It’s time for a factory reset.

Subscribe now

Securing PRA approval

PRA review is overseen by the Office of Information and Regulatory Affairs (OIRA) within the Office of Management and Budget (OMB). The process requires following a highly structured sequence that takes a minimum of six months.

When an agency designs any form or data collection, it must prepare an information collection request (ICR) for multiple layers of internal review. The ICR is a comprehensive package that details and justifies the collection, and must include estimates of the burden on the public and the cost to the government, both clumsy things to compute and based on estimated effort and wages.2 The PRA guidance is helpful, but misleadingly makes it sound trivial to compile.

Once the ICR is approved internally, the agency has to publish a 60-day notice in the Federal Register to solicit public comment. During this phase, the agency is in a holding pattern to allow for input on whether the collection is necessary, understandable, and minimally burdensome. The agency then determines how to handle each comment — which can include grouping similar comments or ignoring specific ones — and must document every decision for OMB review.

After any revisions, there’s another Federal Register notice to invite another round of public comment, though this time for only 30 days. At the same time, the agency submits the full package to OIRA for review. The documentation must include the instruments themselves, the guidance provided to respondents, detailed justification for all decisions made during the comment period, and screenshots or mockups of any system interfaces (often requiring dozens of pages capturing every field, dropdown, upload, and attestation).3 OIRA has up to 60 days to review the submission, but must wait at least 30 to account for any public comments. Upon approval, OIRA issues each collection an OMB Control Number. Without it, an agency cannot legally collect information. The control number typically expires after three years, at which point the agency needs to go through the whole ordeal again.

Let me give you an example of this process in practice. Early on, we knew we’d need to systematically track and manage meeting and engagement requests from the public, applicants, and other government officials. To handle that onslaught, we built the AskCHIPS portal, which helped us gather information upfront and ensured that meetings and speaking engagements were properly vetted and staffed.

The form on the portal is voluntary and takes less than five minutes to complete. Despite getting clearance for an expedited emergency process (more on that below), we still needed to produce:4

• a Federal Register notice;

• a Privacy Act notice explaining how the data would be used;

• a 20-page document with annotated screenshots to show every permutation of the data entry page;

• a 7-page Supporting Statement defending the collection; and

• an emergency justification letter to make the case for an expedited review.

Even though this felt like overkill for a voluntary meeting request form, we had no choice — agencies are not allowed to collect information from ten or more members of the public unless the form displays a valid control number. Without it, the PRA’s public protection provision kicks in: if respondents fail to comply (whether by not filing the form or leaving out fields), agencies cannot penalize non-response or deny a benefit on that basis. Though that legal backstop was largely irrelevant for a voluntary form like this one, the consequences would be much more significant for all of our other collections. That legal risk is the core motivator for bothering with the PRA process at all.

The sequencing problem and how we handled it

The core PRA implementation challenge is sequencing. The standard review process requires managing three distinct and complicated timelines:

• The first is the policy timeline, which is a function of program urgency. Ideally, this would set the pace of the rest of the system. In our case, Congress had issued a mandate, the Department of Commerce had public commitments about when funding opportunities would open, and industry needed us to move quickly — the expectations were clear.

• The second is the product timeline, which needs to allow for iteration and refinement. We were building a digital application system that needed to handle complex, conditional workflows; large financial uploads; multiple projects per applicant; and sensitive information tied to national security and supply chains. Product work is inherently iterative. You test your system, realize that the instructions are confusing, adjust, and learn.

• All of this is frustratingly slowed down by the PRA timeline. The standard review procedure needs to start at least six months before program launch to get approval in time. The process also assumes that you know exactly what information you need to collect and how, and that you can lock the details in early enough to publish for comment and review. The screenshot requirement is particularly annoying, occasionally forcing teams to build wireframes before they’ve had the chance to figure out what the product actually needs to do.5

On the whole, the PRA requires designing your system around the compliance timeline — it’s like allowing your slowest runner to take the longest leg of the relay race. What’s worse, the process almost certainly discourages agencies from collecting information to get the insight they need.

When I arrived at CHIPS in early January 2023, we were careening toward the Notice of Funding Opportunity (NOFO) publication date in late February. As soon as we published the NOFO, we would need to start collecting information via the Statement of Interest, but neither that nor any of the other applicant-facing interfaces were designed yet. We were developing policy, and the portal through which applicants would eventually submit all materials would need to reflect those changes. CHIPS was signed in August 2022; even if we kicked things off immediately (which would have been impossible without leaders or a programmatic vision in place), we’d have been behind.

Because of these constraints, we had no choice but to pursue emergency PRA approvals. The emergency procedure is intended for high-stakes circumstances like ours: the statute allows OMB to approve a collection for up to 180 days if waiting for the standard process would cause “public harm.”6 In emergencies, OMB also has discretion to modify or waive parts of the public comment process. But emergency PRA is ultimately not much of a shortcut. In our case, it still required fulsome packages — Supporting Statements, instruments, screenshots, instructions — and came with an explicit expectation that we would immediately begin the standard clearance process to replace the emergency approval before it expired.

In practice, the emergency PRA process can increase the effort needed to produce a collection. You do everything once to launch, and then repeat the process just to check the box. Although we managed to submit some forms early enough to run the standard process, half of our collections — including the AskCHIPS form — required the “shortcut” and the redo in order to remain in place. This felt like compression rather than flexibility.

Another case of misaligned incentives

The PRA comes with its own institutional ecosystem of staff who are tasked with shepherding collections through the review process. For CHIPS, it was the NIST PRA officer, the Commerce Department’s PRA lead, and the OMB desk officer within OIRA who ultimately approved our collections. Every collection had to move through that chain, in that order.

The incentives of the PRA stewards are often in tension with program urgency. Stewards need to be faithful to the process and to protect their relationships. Each person in the chain wants to ensure that what they send to the next can withstand scrutiny, and OMB must maintain consistency across the federal government. No one is trying to make your life difficult — they’re just preserving the system and their credibility within it. But in effect, even trivial edits must clear a process hellbent on procedural integrity. In our case, everyone was maximally helpful, but the process does not optimize for outcomes.

Institutionalization is a sensible response to volume and complexity. You need to ensure consistency, defensibility, and fidelity to the rules; hiring, procurement, and financial reporting, among other things, all depend on it. But unlike other systems that incentivize process, the value of PRA is seriously in doubt. Hiring and procurement at least promise new capacity. Even if the process sucks, you still stand to benefit. With PRA, the promised benefit is burden reduction for the public and improved information quality. The methods used to evaluate that are inaccurate, and in my experience, PRA most zealously protects process itself — proper notice, proper documentation, proper sequencing, proper review. I do not recall getting any meaningful feedback on any of our collections from anyone in the review chain.

The PRA might spare the public of burdensome paperwork, but it imposes hidden costs on the taxpayer. The staff having to navigate the PRA procedure could have applied their time and capacity to far more strategic efforts; the senior folks tasked with overseeing the process would be put to much better use designing the investment strategy, negotiating awards, or supporting applicants. The PRA did nothing to improve our outcomes but it did drain significant mindshare from the program during its early days. And ultimately the public pays for it. We managed to achieve our goals, but who knows what else we could have done with all of that misdirected effort?

Subscribe now

It’s time for a factory reset on the PRA

The spirit of the PRA is defensible — it’s good for the government to minimize unnecessary burden where possible. But the law and process need an update. A wholesale reimagining could preserve the spirit and improve the machinery, allowing public servants to focus on what actually matters.

There has been underwhelmingly limited study of the value of the PRA, but it’s already clear that it falls short of its aims. Reports by the Government Accountability Office (GAO) and the Administrative Conference of the United States call into question the utility of the redundant comment periods. The GAO report found that only a small fraction of collections received any comments at all, and that the resulting changes were typically clarifications rather than substantive changes or reductions in burden. This was true in our case as well: we received mostly clarifying questions on the application materials and only a handful of substantive comments across all of our collections.7 If the point of the comment period is to understand and reduce burden, that process is demonstrably not delivering.

GAO has also documented the brittleness of the burden estimates themselves. Per the report, one ICR underestimated burden cost by as much as $270 million, while another overestimated burden time by over 12 million hours. It comes as no surprise that agencies can’t reliably estimate something so abstract — our own burden and cost estimates were inevitably a shot in the dark and likely wrong. The one metric of PRA’s efficacy is impossible to measure accurately.

At the very least, it’s obvious that we need to either pare down the redundant comment periods, or remove them altogether in all but the most sensitive or complex collections. In general, the procedure should distinguish between low-risk, low-burden collections and high-burden, high-stakes reporting or statistical regimes. Other process improvements could include delegating authority to agencies and not penalizing system fixes or redesigns that enhance user experience but do not come with substantive changes.

Recent years have seen limited attempts at reforming the PRA, including the creation of generic clearances for customer experience and service delivery feedback, which allow agencies to operate under a pre-approved umbrella rather than seeking full PRA clearance for each individual collection. But these reforms are too narrow. CHIPS wasn’t trying to run a customer satisfaction survey. We needed to stand up a national industrial policy program with a digital application system and responsible oversight. The PRA gets in the way where the government most needs speed, competence, and insight. And I suspect that there’s an added cost of agencies deciding to not collect data and information that could meaningfully improve a program’s outcomes, just to avoid the PRA process.

Statute, regulations, and agency norms would all need to be rethought to break the PRA paradox. In the 30 years since the last revision of the PRA, it has arguably failed to reduce burden (indeed, burden has increased in that time!). The PRA’s failure in turn threatens the success of the government initiatives subjected to the process.

From my perspective, the paradox is especially pronounced in a program built at the speed and scale of CHIPS. Across my two years at CPO, there were 25 days (five full work weeks!) where we were dealing with some form of formal PRA transaction — and that of course doesn’t account for the hundreds of person-hours that went into generating the required documentation and managing the process. We got better at managing the cycles as the program stabilized, but there was undoubtedly a cost to our team capacity, especially in the early days, when we could least afford it.

Commonplace

A magazine about what matters in America

By Oren Cass

The CHIPS Act was billed as a once-in-a-generation effort to rebuild America’s manufacturing base in a strategically vital industry. But turning legislation into functioning factories and good paying jobs requires far more than slogans about “onshoring” or wish-casting. It demands a state-sponsored investment outside of America’s typical comfort zone.

Mike Schmidt, former director of the CHIPS Program Office and co-author of Factory Settings, joins Oren Cass, Founder and Chief Economist at American Compass, to discuss what it actually took to stand up the largest industrial policy initiative in decades. They explore how the government negotiated with global chipmakers, why grants and tax credits were combined, what critics missed in the “everything bagel” debate, and how permitting, labor, and geopolitical risk shaped their efforts. They close by discussing how we’ll know if the CHIPS Act ultimately succeeds and the way the US should think about future reindustrialization efforts.

“Victorious warriors win first, and then go to war.” Sun Tzu

Policy work sits at the top of Washington’s prestige hierarchy. Shaping ideas and legislation is held in high esteem; by contrast, deciding how to actually implement a vision is undervalued. And execution talent — people who can convert policy into real programs — is in short supply.

To build the CHIPS program, we first needed to translate our statutory mandate into a coherent, implementable strategy. Intentional program design wasn’t the entirety of execution, but success would have been impossible without it.

The CHIPS Act did not specify execution strategy

The CHIPS Act was signed into law on August 9, 2022. When Mike and I joined the team right after Labor Day, we were still a ramshackle start-up troupe of mostly “detailed” talent borrowed from other bureaus or agencies. And we had $39 billion burning a hole in our pocket, with high expectations from both the White House and one of the most sophisticated and politically savvy industries in the world.

Section 9902 of the legislation offered a mandate and a high-level mission, instructing the Secretary of Commerce to create a financial incentive program to strengthen national security by building domestic semiconductor capacity. The specifics — strategy, segmentation, prioritization, and sizing — were left almost entirely to the implementing team to determine. For example:

• CHIPS authorized $39 billion across grants, cooperative agreements, other transactions, loans, and loan guarantees (up to $6 billion of which could be used as subsidy cost to support up to $75 billion of loans and loan guarantees) — but it did not specify the balance of each or when a particular tool or strategy might be better suited to the task.

• The legislation identified eligible uses of funds (e.g., construction, expansion, or modernization of commercial semiconductor fabs and related supply chain facilities) and required at least $2 billion to be allocated to legacy or mature node capacity essential to national security, but did not recommend specific allocations and prioritization beyond that (i.e., how much should go to leading edge logic, and how to prioritize funding memory vs. logic, or supply chain vs. fabs).

• CHIPS laid out broad prioritization criteria — projects needed to strengthen supply chains, advance national security, promote economic competitiveness, and demonstrate financial viability — but did not provide further selection logic or sizing direction.

Our job was to turn this enabling legislation into an executable industrial policy program. There were hundreds of design, process, and organizational structure decisions to make: What should we do first? How will we size individual awards? Should we prioritize loans? What is most important for national security (and how do we even define that)? Building a capable program required quickly identifying which decisions would prove most significant.

Subscribe now

Deciding how to execute

Four early decisions shaped how we executed the CHIPS program and ultimately set us up for success. Rather than a reactive, “let’s see what comes in” approach, we set our intentions and expectations early and codified them in our Notice of Funding Opportunity (NOFO).

1. Using rolling applications

One of the first design decisions we had to make was whether to create one universal deadline for all applications or to evaluate them on a rolling basis. Many grant programs choose the former — it’s easier to organize a review process around a fixed date, compare all applications against each other, and announce all awards together.

Program statute often sets a short-term date by which all money must be allocated, making the universal deadline a more efficient option. But this model gives an edge to the most organized and government-savvy companies and to projects that are furthest along in development (which would likely materialize even without our funding); it’s harder for nascent ideas and technologies to compete with more detailed, mature ideas in this process. A batched rolling process also constrains resources and timelines, making it challenging to work with applicants to shape projects.

CHIPS did not define a short-term target date. This was to our benefit — we knew that the industry would evolve over the coming years (ChatGPT was not even launched when the legislation was passed) and that many important industry players — especially foreign companies — were more likely to come to the table over time rather than respond to a tight timeline. We wanted to allow and encourage companies to apply as soon as they were ready. We also wanted the ability to proactively shape the end result — rather than fund what was already in a company’s plan, we wanted to push them to pursue more ambitious projects, which would take more time to develop. And we wanted to leave room for emerging technologies, such advanced packaging and glass substrates, which we expected would need additional time to translate into well-defined, fundable projects. Rolling applications just made more sense.

But a rolling process creates its own set of challenges. We needed to manage resources without set timelines and to equip our team to proactively push reluctant companies. And, most critically, we had to make sound decisions without the full picture of what might come in.

2. Proactively slicing up the pie

$39 billion sounds like a lot of money — until you put it into context. TSMC alone was expected to spend over $40 billion in capital expenditures in 2025. Global semiconductor revenues in 2024 exceeded $600 billion; industry capital expenditures were over $150 billion. We were focused on funding capacity buildout through 2030; $39 billion spread over those six years is merely 1% of current global annual revenue and 4% of global capex. We clearly wouldn’t be able to do everything we wanted to, and we needed our dollars to go as far as possible.

Opting for a rolling application process complicated things. With a fixed application deadline, you can review the entire set of applications and select the strongest ones, allocating funding accordingly. But the combination of a rolling program and limited capital in a highly dynamic industry meant we had to make high-stakes decisions without knowing our complete portfolio. Our four leading edge applicants alone requested $70+ billion in total funding — nearly twice our total budget. Without a plan we risked overallocating to early applications and not being able to fund the reluctant applicants or emerging technologies that came along later in the process.

To mitigate that risk and guide our outreach and decisions, we proactively outlined a target allocation of capital across the various industry segments. Doing so required documenting what we could realistically accomplish and what our ideal portfolio would look like. We broke this down further: How many leading edge fabs would we ideally fund? How much memory capacity? How many packaging facilities? How much for the upstream supply chain? Luckily, we had the investment tax credit at our disposal, which applied to all semiconductor projects and we thought would be sufficient incentive for some projects. We needed to focus our dollars on the critical projects that the tax credit alone would not catalyze.

We had articulated our high-level goals and priorities in our Vision for Success, but we needed much more specificity to truly guide our overall effort. For example, the document committed us to at least two new leading edge logic clusters, but we still needed to define what a cluster was, how many fabs it would consist of, and how much to invest while still leaving enough money for our other priorities.

Our strategy and investment teams dug into the economics of each subsegment, turning to outside consultants to gain deeper industry insight. We started with supply and demand — we broke down the industry by node, modeled all current and announced supply, and evaluated multiple detailed demand projections. We studied past industry cycles and grappled with questions about how AI could shift demand growth and how oversupply in China might impact different legacy nodes. This helped us determine how many new fabs would be needed, which made a clearer case for convincing companies to build them in the US instead of elsewhere.

But that was only the beginning. We then built cost models for different types of fabs and used proprietary industry data to estimate cost differentials with other regions. We had to understand the scale of investment needed to accomplish our detailed goals, what it would cost in the US compared to other geographies, and the returns that companies would require to commit to building in America. And these economics varied significantly by sector. A typical leading edge fab costs north of $20 billion; TSMC, the industry leader, generates operating margins over 45% company-wide. A mature node fab would cost about half of that amount, but with much lower margins (GlobalFoundries operating margins are in the mid-teens). Contrast those with advanced packaging facilities, which may require only $2 billion of investment, but where overall company margins have historically been in the single digits.

Quantifying the expected costs of the priorities outlined in our Vision for Success helped us determine funding needs across the ecosystem. In estimating the range of new leading edge fabs required to meet global demand through 2030, we determined that roughly 8-10 of them would need to be in the US to reach our target 20% market share.1 While possible, it would require $200+ billion, so we needed to be judicious about how much of our funding to direct at that goal.

Our internal roadmap sharpened our sense of what success could look like and how much capital to allocate to each sector. It also allowed us to shape our applicant pipeline rather than react to it, and to evaluate applications with a clear goal in mind. We were not bound to hard targets — we knew we would have to adjust based on the actual applications we’d receive and how the industry evolved. But the roadmap clarified our goals and established discipline and focus. We iterated on our allocation targets as we received new information, reevaluating our initial analysis with the investment committee and the broader team and adjusting the portfolio accordingly.

3. Sizing awards

With our target portfolio allocation established, we had to decide how to size individual awards and what to communicate upfront. We believed that setting expectations early would boost our negotiating leverage and help anchor our approach to applicant engagement.

Our mantra on sizing awards was simple: give applicants just enough subsidy — and not a penny more. We wanted to catalyze private investment, not displace it. And we needed a consistent framework by which to evaluate applications.

First, we established a strategy for award sizing. We borrowed a page from corporate finance and private equity, implementing the internal rate of return (IRR) vs. hurdle rate methodology taught in every finance course in the world. This strategy — which many companies and investors use to make investment decisions — compares long-term annualized returns on invested capital against a company’s historical cost of capital. This calculus was familiar to companies and grounded our negotiations in numbers rather than anecdotes or vibes. It wasn’t a pure science — factors like national security importance, a company’s other options, and project risk all factored in — but it created a strong starting point. As the NOFO put it:

“…the degree to which the request for CHIPS Incentives is reasonable and necessary to make the project viable … [will be] based on cash flow modeling, IRR analysis, sensitivity analysis, and other applicable analyses, and whether the projected IRR in the cash flow model is reasonable based on the expected risks and returns of the project, historical projects of similar nature, or other relevant market benchmarks.”

We didn’t set a specific IRR target. Instead, we crafted a framework aimed at producing a return attractive enough for firms to build in the US. We paired it with another critical clause:

“Most direct funding awards are generally expected to range between 5-15% of project capital expenditures.”

Fifteen words in a 129-page, 75,000-word document — yet perhaps the most consequential and risky. We could have chosen to omit this range; there was a chance that these relatively low percentages might discourage companies, especially those on the fence. And what if we couldn’t hold the line in negotiations — were we setting ourselves up to fail?

But we didn’t pluck that range from thin air; our team — including our excellent Chief Risk Officer, Andy Kuritzkes, and a team from Oliver Wyman — conducted a detailed analysis to determine what “just enough” subsidy to generate attractive returns might be. To establish a realistic range, they modeled sample returns for different types of projects based on historical project investment returns, global cost differentials, and detailed operational models at different subsidy levels.

Our analysis affirmed that the 25% tax credit was a meaningful baseline incentive and, in some cases, sufficient to catalyze investment (especially when paired with state or local incentives). Our funds would therefore be better used for important projects that required an incremental push. The 5-15% range also meant that public support would still be below half of total capex (25% tax credit, plus our 5-15%, plus state and local contributions, which are typically quite modest). The companies themselves and other private capital would be the majority investors with real skin in the game, aligning their incentives with ours. And since we tied support to upfront capex and not ongoing operational costs, companies were incentivized to find creative ways to shrink ongoing costs over time. Lastly, establishing a range also conceded that financial viability varied across subsectors; an advanced packaging facility with thinner margins would likely require a higher percentage of support than a leading edge fab.

We could have forgone the complicated IRR methodology and set a consistent target percentage for all awards. This would have considerably simplified our process and created less oversight risk, but at the expense of flexibility and the ability to shape ultimate outcomes through differentiated incentives. Establishing and communicating our approach early set expectations and bounded negotiations. And it positioned us as a serious counterparty grounded in analytical rigor, redirecting the conversation from the government affairs teams of these companies to their operations and finance executives (a subtle, but important distinction).

Defining our funding range also helped with our portfolio math — we knew that if we funded at our 10% midpoint on average, our $39 billion could catalyze nearly $400 billion in total investment. Importantly, it also created an analytical framework and discipline for our internal teams and investment committee that proved valuable when debating and defending sizing decisions.

4. Closing information asymmetry

I had a lot of anxiety about taking on this job. We were handicapped by enormous disadvantages in negotiations: the companies on the other side of the table knew far more about their cost structures, had armies of well-paid analysts, could hire the best advisors, and had substantial political influence. How were we not going to get eaten alive?

We needed to level the playing field. To signal our seriousness, we hired top talent. We also developed our understanding of the industry through iterative applicant engagement and built trusting and open relationships with customers and investors. And we borrowed another page from private equity, striving to align interests and force transparency.

Our sizing approach played a part here, since it required companies to disclose data and contribute their own considerable investment. Even more significant, though, was our decision to add an upside-sharing provision to the NOFO:

“Recipients of more than $150 million in direct funding must share with the U.S. Government a portion of any cash flows or returns that exceed the applicant’s projections above an established threshold. The terms will be set on a case-by-case basis and are intended to incentivize accurate projections and ensure that taxpayers benefit from outsized financial returns” [emphasis added].

This clause was not required and drew significant industry criticism.2 And it sparked intense debate within our team and with the White House — did we need to complicate the program if this was truly about national security outcomes?

While our upside-sharing mechanism was designed to benefit taxpayers, we didn’t expect to get much money back. The clause’s biggest near-term benefit was that it reduced information asymmetries and incentivized companies to provide accurate projections. Upside sharing discouraged companies from gaming the system by giving us a conservative financial case to justify more subsidy — outsized profit would just come back to the government.

Our award sizing methodology undoubtedly created complexity, but it made clear that this was not your typical government grant program. Together, our sizing and upside-sharing clauses reduced the company’s edge and aligned their interests with ours.

CHIPS gave us discretion

The CHIPS legislation defined the program’s high level goals, but not how to go about achieving them. Our program design was geared toward keeping us on our front foot — effective execution boiled down to tactical early decisions that drove internal focus and gave us more leverage in negotiations. By establishing a proactive allocation framework, using rolling applications, and defining a transparent sizing and upside-sharing methodology, we set expectations for companies, provided clear direction to our team, and leveled power and information asymmetries.

These were carefully considered decisions on how to execute on a broad mandate. Our experience affirmed why the corporate world believes “execution eats strategy for breakfast” — government needs to internalize this maxim and place more emphasis and value on program design and execution.

We hope you have been having a restful holiday season.

We are excited to republish an interview our colleague Santi Ruiz at Statecraft conducted with Gerry Petrella, a senior aide to Senator Chuck Schumer (D-N.Y.) during the passing of the CHIPS and Science Act. We hope that in light of Factory Settings’ recent writing on execution readers will have a newfound appreciation for the legislative negotiations behind the effort involved.

We look forward to resuming our standard publication schedule in 2026. Until then, we wish all our readers a happy new year!

Please read on for a transcript of the interview with Statecraft’s Santi Ruiz. We encourage you to subscribe to Statecraft at the link below.

Statecraft

How policymakers get things done

By Santi Ruiz

In 2022, the U.S. passed the Creating Helpful Incentives to Produce Semiconductors (CHIPS) and Science Act, which authorized $280 billion of investment in a broad range of targets, including quantum computing, clean energy initiatives, NASA, semiconductor manufacturing, and $24 billion in tax credits incentivizing domestic semiconductor production to diminish supply chain reliance on China.

CHIPS and Science combined two bipartisan Senate bills: the Endless Frontier Act and the United States Innovation and Competition Act (USICA), in a complicated negotiation that involved both chambers of Congress and a rarely-used conference process.

In March 2023, Senate Majority Leader Chuck Schumer said on the Senate floor that both the CHIPS and Science Act and the Inflation Reduction Act “wouldn’t have happened without Gerry Petrella.” Petrella, a former longtime aide and policy director for Schumer, had worked for months behind the scenes to build the support needed for both bills.

Statecraft talked to Petrella about how the CHIPS negotiation went down.

What You’ll Learn

• What are the functions of the majority leader’s staff?

• How does information flow in the Senate?

• Do committees still work, or has all negotiation moved to backrooms?

How does the office of the Senate majority leader work differently from the office of a run-of-the-mill senator? What are the organizational differences?

There are a couple of differences. First, there’s inherently a lot more decision-making that has to happen to actually run the Senate, because the mechanics and scheduling of the Senate are in the domain of the majority and minority leaders. I operated on both sides of that coin for Leader Schumer.

Because the Senate is a body that has to work, more often than not, by unanimous consent. Even waiving the morning business, which can take hours, requires all 100 senators to agree.

Senate operations require the two leaders, the staffs they have on the Senate floor, and the political and policy staff behind them to work together on both micro and macro level decisions. Together, they decide which legislation we will actually consider in the Senate, and when and how it goes from just being a bill or an idea, to a bill coming out of a committee, to getting to the Senate floor. Both leaders have to be a part of that process, and usually very early on.

The majority leader staff has a larger apparatus. By the numbers, the office is maybe double the size of a senator’s personal office. The staff includes multiple domain experts — health care, transportation, environment, and so on — to field incoming requests, answer questions, and provide leadership, vision, and strategy. The office also helps the committees and each of the senators’ offices, especially those on your side of the aisle, understand what’s going on, problem-solve, and take the incoming. There’s a lot of decision-making that has to happen there, so there’s more staff and more involvement.

I worked in the personal office for Schumer before he was leader, and it’s interesting to see the differences. There are certain things you wonder about when you’re not in the majority leader or minority leader’s office, like, “Why’d they make that decision? Why are we doing this? What’s going on?” Once you’re actually there, you’re like, oh, now I understand.

How does information enter the majority leader’s office, and what does that information diet look like? How does it get synthesized?

It’s really a tremendous effort on the part of so many people, including not only the leader’s office, but also the whip’s office, the chair of the policy and communications committee, and the press and speechwriting offices of the leader.

There are teams of people dedicated to getting information out to senators. When the leader gives remarks on the floor, someone has to write those. We process tons of complex parliamentary information every day. It often doesn’t sound like English. So the whip’s office and floor staff will translate that through emails and other forms of communication. There’s a broad apparatus dedicated to just doing that.

Go a level deeper for me on the distinction between floor staff and policy staff. What are the roles and day-to-days of the floor staff?

In the Senate, both the majority and the minority leaders appoint their party secretary. There are not many roles that the senators themselves appoint, but Senate rules let them select the sergeant at arms and their secretaries.

You’ll see the two secretaries, currently Gary Myrick and Robert Duncan, are always on the Senate floor, and they wear a special pin. They are in their own league, really, because the entire Senate has granted them the privilege of helping make decisions about parliamentary procedure.

Those two secretaries each have a staff, colloquially called the floor staff. They’re wonderful, brilliant people who work way too many hours because the Senate often does business late into the night on Monday, Tuesday, Wednesday, and Thursday nights. They keep the place functioning.

Under the two secretaries are usually a deputy and a few other staffers who handle particular aspects of the parliamentary process for the Senate and senators. They help introduce bills, file amendments, and prepare scripts for leaders and senators to read to conduct floor business, but are separate from the policy staff in the party leader’s office. I worked very closely with Gary Myrick, who is an unbelievable individual and has been there for many years through many leaders, but our roles were different. My job was to figure out how am I getting a bill, a substantive piece of legislation to the Senate floor for it to be successful?

When it comes to bills like CHIPS and Science, how much of the nitty gritty is driven by senators themselves, vs. by staff?

That’s a great question, and it’s relevant to not just CHIPS and Science, but to the legislative process in general. The Senators are the key decision makers and visionaries for legislation – they set the path and call the final shots. However, I think senators, rightfully and smartly, rely a lot on staff — not just their own staff, but the staffs of the leaders and of the committees. It’s almost too complex and too much volume for any one human being to handle.

My former boss Leader Schumer relied on his team to help him execute the vision and goals he and the conference had. We helped with getting bills, ideas, and issues to a place where they could come out of the committees with bipartisan support and then come to the Senate floor. Staffers also help negotiate some of the more complex terms and policy issues inside of bills, like the Inflation Reduction Act and CHIPS and Science.

I think senators, writ large, do that. Let’s use CHIPS and Science as an example. Almost every senator has an aide who handles technology, environment, and climate change policy. Some offices break it up differently by discipline, but almost all of them have staffers who handle technology and science issues.

But then the Commerce Committee has its own suite of professionals who report to the Chairwoman and to the ranking member. Those folks are very technically savvy, and they are experts in actually writing statute. Oftentimes, senators will bring their ideas or bills to the Committee staff and say, “Here’s my idea,” or, “Here’s how I drafted the bill,” and then encounter the committee staff whose domain this really is. A bill’s outcome sometimes depends on whether the Senator wants to listen to the committee. The committee will want to put their imprimatur on the legislation, or correct how you’ve amended the statutes in their jurisdiction. I would always advise listening to the Committee professionals if you want to get a bill passed.

The Senate staffs are really the experts in many issue areas, and they help bring the senators’ ideas to life and move them through the process. The senators help provide vision, clarity, and strategy. Negotiation with other senators, especially on tricky issues, also often gets boiled up to their level. Let’s say there are 100 unresolved issues at the start of a negotiation. The staffs will try to negotiate as many of those issues as possible, but inevitably there will be two, three, or ten that they need to, as folks would sometimes joke, kick up to people with election certificates.

When you have bipartisan support, as you did with CHIPS and Science, how often do senators have to be in the room to hash out the details?

CHIPS and Science required a lot, because it was negotiated over the course of almost two years. Schumer himself was the original sponsor of the Endless Frontier Act, which was one of the individual pieces of the final bill. He also helped negotiate the original terms of it with Senator Young and others, so they were very, very involved.

The bill then went through a committee process where Chair Cantwell, ranking member Wicker, and others and their staffs said, “We didn’t like exactly how you did it, so we’re going to try to do it our way.” I was on some late-night phone calls with Senator Cantwell herself right before the markup of Endless Frontiers Plus, which wound up being called the U.S. Innovation and Competition Act (USICA) in the Senate, and then was signed as CHIPS and Science.

There were huge dramas related to some of the language that senators themselves had to address, like the Davis-Bacon language in the CHIPS program. Senators Stabenow and Warner were involved, for example, and they weren’t even on committee. There are always these moments when you’re handling major legislation, when the senators themselves are very involved as it goes through the committee markup process.

Like the amendment process, committee markups are held live and the senators are present. The staffs help them organize and make decisions, but the senators are the vessels to take the votes, give the speeches, and actually pass legislation.

It’s a symbiotic relationship. The staffer only exists because the senator has offered them the pleasure of serving, and the senator really needs the staffer in order to do the job.

When a bill like CHIPS and Science passes, how much of that legislative success do you attribute to careful procedural and tactical work on the floor and committee wheeling and dealing?

CHIPS and Science is an interesting example because it had fits and starts and looked like it was dead for a while. That one maybe more than any other examples of legislation, I should say. It’s pretty unique, because it started out in the Senate as a large, comprehensive, and bipartisan piece of legislation. The House really didn’t have input, and they reminded us of that very often in the first year of that Congress. It was contentious.

Speaker Pelosi and House Democrats were not that happy with the way we did it, so the House drafted and passed their own version of the bill, called Creating Opportunities for Manufacturing, Pre-Eminence in Technology, and Economic Strength (COMPETES). We tried doing an informal conference between the two chambers, and while we were pursuing and passing this bipartisan legislation with the Republicans at the time, we were also doing some partisan legislating behind the scenes, which wound up becoming the Inflation Reduction Act.

It was hard to get a real negotiation going with the Republicans on CHIPS and Science because they were using it as leverage over the Inflation Reduction Act. We tried negotiating informally with the House, but that didn’t go anywhere.

To answer your question about whether there were any unique parliamentary moves used, we threatened to amend our Senate National Defense Authorization Act with the Senate-passed version of CHIPS and Science. It was getting close to the end of 2021, and we were frustrated that we hadn’t made any progress in our conference negotiation with the House. We were playing what’s called a ping pong, which is a legislative term for when the House and the Senate amend one another’s bills to try to get to a final product, rather than go through the conference process.

We got the idea to do that on the National Defense Authorization Act (NDAA) for 2022. Leader Schumer told me to go do it. I thought, “If this doesn’t work, let’s at least try to shake something loose.” So we threatened to put our version of CHIPS and Science on NDAA.

The Senate Republicans and House Democrats didn’t like that, so we were able to extract a commitment to do an official conference.

To clarify for our readers, the threat that you were making was that you would put your version of CHIPS and Science into legislation that had to pass.

Or at least that it then becomes a lot harder to get your own things in, because we were attaching it to the defense bill, yes. But out of that came a commitment from all of the leaders and the chairs of the relevant committees to form an official conference. Congress rarely ever does that anymore, but it is a real, technical vehicle for resolving issues between the two chambers.

Why are official conferences so rare?

You know... it’s interesting. You can’t do anything in Congress easily anymore, and even creating the mechanisms to resolve problems takes time. Just debating the motion to go to conference in the Senate, without cooperation from the other side, can take days or weeks, and you’re not even doing anything substantive. That’s one reason why it doesn’t happen often anymore.

Another reason, which some people bemoan, is the centralization of decision-making behind closed doors. Because of that first reason, among others, the committees and the leaders just negotiate on a bipartisan basis and resolve their issues informally. It’s essentially a conference negotiation without forming the technical committee.

I’m a nerd for procedure, so when we formed the technical conference committee to reconcile the differences between USICA and COMPETES, I thought, “This is cool. We haven’t done this in a while, and we made it happen.” The conference committee is required to have public meetings, so the senators and the House members who were appointed by the leaders on both sides all showed up and talked about how much they wanted to do it, but we still couldn’t get the thing going in the formal process.

Fast-forwarding to the waning days of Congress in the summer of 2022, the clock is ticking, we’re trying to pass Build Back Better, and negotiating with Manchin. We passed the bipartisan infrastructure bill and the veterans bills, we had Ukraine pop up and were doing tons of bipartisan stuff, but this one budgetary thing we were going to do on a partisan basis was still hanging out there and CHIPS and Science, not born of that name yet, was still hanging out there.

We hadn’t fully resolved all of our issues with the House, but because we created the conference committee and the conversations had been going on for so long, an understanding formed of what often is referred to in the halls of Congress as the “lowest common denominator.” In Congress, that means, “We can’t resolve all of the issues we’re fighting about, but what can we all agree to on a bipartisan basis? We can’t resolve all of the issues we’re fighting about, but what’s the lowest common denominator?”

Slowly it became clear that the trade provisions and some of the immigration and research restrictions in the Senate bill were just too controversial for the House Democrats, and we couldn’t resolve them with the remaining time. Through discussions, we understood that everyone wants the semiconductors, the chips part was hugely bipartisan, and everyone really liked the science pieces. We had to make some changes to accommodate the House, but we thought that maybe we could just go by on the chips and science parts at the end of the day. If we had to go, maybe we could just go on that. And that conversation began behind closed doors and not in a formal process.

Was that stance from the leader’s office purely in the interest of moving things through in that term of Congress?

Yes, we had to move things through. We thought, we have an election coming up in three months, the window is closing to legislate on something substantively important, and our members want to run on something.

The Senate ultimately passed CHIPS and Science in mid-July, before the August recess. It then went over to the House, which Speaker Pelosi controlled, so we expected to have no problem there.

What did the fights over CHIPS and Science implementation focus on? Was it about how the money would be disbursed?

I was kind of on my way out at that time, so I started the process of managing the implementation of those bills, but I didn’t get to see the fruition of it. I did just see that all the tech hubs got announced.

CHIPS and Science is a tricky one because you have limited resources, but the outcomes are very important for the country, and for individual states and members, so there’s a lot of jockeying. Where are the fabs going to go? Where are the hubs going to go?

That’s fine. Before we wrap up, are there other critical aspects of the legislative process that we haven’t discussed?

I’ll just wax poetic for a bit because I’m proud of this and I fight with Paul Kane at the Washington Post about this often. For a long time, or at least in the 15 years I was there, people have said that all of the power is with the four leaders and the White House and that all of the deals get cut behind closed doors and in back rooms, that the committees are powerless and don’t do anything.

Leadership does have to negotiate the terms, but much of the transformative legislation, like CHIPS and Science, the bipartisan infrastructure law, the CARES Act, COVID bills, gun bills, and much of the transformative stuff in the Inflation Reduction Act, is written by the committees. Sure, the leadership negotiated the big terms. But the committees write all that stuff. And they often wind up negotiating with Republican ranking members and members down dais [nb: junior members of Congress].

The CHIPS and Science Act was all legislation that got reported out of the committees, and the bipartisan infrastructure law was like seven Senate committee bills put into one. The leader’s office helps provide vision and negotiates and problem solves, but I don’t think it’s fair to say that the House and Senate committees are not critical. They’re incredibly important. The leaders don’t write bills out of thin air. They don’t have the capacity and that’s not the way the system is set up.

I have this long-standing point that I often make which is that the Senate is not an atrophied institution. It still works pretty well when you let it.

Are there procedural changes that you would make to the Senate as it exists today?

Yes. Yes. I would update the rules to allow more efficiencies. People complain, “Why don’t they do the 12 appropriations bills?” They just tried to pass three of them, and it took seven weeks. There should be special procedures for certain types of bills. If you have overwhelmingly bipartisan support, you should be able to move faster.

Thanks to Rita Sokolova for her judicious edits on this interview.

We hope you have been enjoying Factory Settings!

For this holiday season we have decided to give you something a little bit different from our usual output.

Today we are crossposting a recent interview we did with our Institute for Progress colleagues at Statecraft. In this episode, we discussed running the CHIPS program office, the challenge of “The Everything Bagel,” the levers we had to disburse $39 billion in program funds, and so much more.

We look forward to returning after the holidays on our regular schedule. Readers can look forward to pieces from Todd Fisher on effective industrial policy execution, Sara Meyers on the Paperwork Reduction Act, and Mike Schmidt on the program’s leading causes of delay. We’ll also have guest posts from writers, policymakers, and other thinkers from across the ecosystem.

Please read on for a transcript of the interview with Statecraft’s Santi Ruiz. We encourage you to subscribe to Statecraft at the link below.

Statecraft

How policymakers get things done

By Santi Ruiz

In the last few decades, few government interventions have been regarded as successes on both sides of the political aisle. Operation Warp Speed, which accelerated the development of COVID vaccines, is one. CHIPS, the Creating Helpful Incentives to Produce Semiconductors Act, is another. It spurred a massive investment boom in semiconductors on American soil. This was led by the CHIPS Program Office (CPO), at the Department of Commerce. The CPO had to decide how to allocate $39 billion of manufacturing incentives — and then negotiate the details with some of the biggest companies in the world.

Today, I’m lucky to have three of the founding members of the CHIPS Program Office team with me:

• Mike Schmidt, the inaugural Director,

• Todd Fisher, the Chief Investment Officer, and;

• Sara Meyers, Chief of Staff and Chief Operating Officer.

Mike, Todd, and Sara have a clear sense of what went right for them, what went wrong, and what they’d do differently the next time. They’re working on a big project for IFP called Factory Settings, in which they describe what they learned.

Subscribe to Factory Settings!

Thanks to Katerina Barton and Harry Fletcher-Wood for their support on this episode.

When you entered the CHIPS office, what did you have to work with?

Mike Schmidt: A bill passes, and what you’re talking about is a startup. There’s nothing — no people, no processes, no strategy. I was the first employee of the CHIPS team. I started in September 2022, about a month after the bill passed. On my first day of work, Secretary Gina Raimondo introduced me to Todd Fisher, who would become our Chief Investment Officer. The key imperatives were:

• Building a vision: how are you going to measure success?

• Building a team: we knew we needed an extraordinary group of people to help us get it done.

• Building a program: the nuts and bolts of how it’s going to work.

The early days were characterized by trying to figure those things out.

Sara Myers: To design a program is to design hundreds of processes from scratch. We had a couple of things going for us in the beginning. We had 30 detailees from across the Commerce Department, trying to help us build the thing. We also had the attention of the senior operations and administrative folks around the Commerce Department, who understood that this was coming and that we were going to need to ramp quickly.

Todd Fisher: The Secretary [had a] sense of urgency and a willingness to do unnatural things, including helping to recruit people — that was critical.

Secretary Raimondo helped you recruit talent?

Todd: This was obviously one of the biggest priorities for the Biden administration; one of, if not the, biggest priority for Secretary Raimondo. She viewed her role — and played it exceptionally well — as helping to break through barriers for us and be the lead recruiter. And the way the statute was created allowed us some flexibility in how we designed the program.

Mike: The other thing we had was $780 million of administrative funding. Most people operating in government are hamstrung by budget. We expected to have to live within that budget for 20-30 years in total, and the spending would be more early on as we ramped up and allocated the funds. But in those early days, we knew we had the resources we needed to make it happen.

Sara: I remember interviewing with you and being like, “So what do we have, resources-wise here?” You were like, “Just don’t worry about it.” I was like, “It’s a thing that you always must worry about in my experience. So what are you talking about?” When you said $780 million, I was like, “He’s not right. He doesn’t know what he’s talking about.”

2% of the total money allocated for CHIPS was for your administrative operations, which is not a huge percentage; but it’s more than enough to run an office.

How do you know CHIPS was a success?

Mike: CHIPS was passed because semiconductor manufacturing capacity had declined pretty dramatically over the course of a few decades, from around 40% to around 10%. Semiconductors are foundational to the entire modern world. On the leading edge — the most advanced chips — production declined to zero. Those are the technologies that are most critical for AI, and the technology and geopolitics of the future.

At leading-edge logic, which is a critical aspect of the supply chain, all the production is happening in Taiwan — a geography of intense geopolitical sensitivity. CHIPS is a bipartisan statement saying, “We have to reverse that trend. That is not sustainable from a national security standpoint. We need to onshore production of semiconductors, in particular leading-edge semiconductors.”

In terms of measuring success, one is we got it done: we allocated the funds. It’s very hard to execute in government. In less than two and a half years, we awarded $34 billion, starting from nothing. That stands up to parallels in the private-equity world.

Towards the end of the program, we were meeting with the Secretary weekly to go through the pipeline, deal by deal. She was turning the screws, as she should have, to make sure we were going to get it done — but also offering to help: “What CEO can I call? How can I unlock this to help you guys get it done?” We were walking out of her office one day — we knew we had 15 or 20 deals to close in the next few months. Todd worked at KKR, a renowned private-equity fund. I said, “Todd, how many big deals do you think KKR will close between now and the end of the year?” He was like, “Two or three.”

Todd: I think people don’t understand what it takes to get a significant deal done. In the private-equity world, maybe a firm would get five or six transactions done in a year. But remember, all of the processes are in place: they have lawyers; they know all the terms that need to be agreed. We had to get the whole process out there: the application process, a portal, get applications in, evaluate and negotiate them — and get a final, signed agreement in place. Not to mention hiring 180 people, getting them integrated, and creating a team spirit. All those things are taken for granted. There’s a reason you don’t give $39 billion to a startup, because it’s complicated to get it up and running.

Mike: The other obvious measure of success: awarding funds doesn’t matter unless that’s having the outcome you want. We’ve seen a massive amount of private investment catalyzed in semiconductor manufacturing here in the US: over $600 billion in announced investments. There are five companies in the world that can produce leading-edge chips: TSMC, Samsung, Intel, Micron, and SK Hynix. All five of them are expanding here. There’s no other place in the world that has more than two of those companies.

The scale of each individual investment is extraordinary. TSMC’s investment is the largest foreign direct investment in the history of the country. Now it’s $165 billion, but the $65 billion deal we signed with them was the largest. They’ve announced the full plan to build out their site there, and plans to build out a second site. TSMC is producing leading-edge chips in the United States — the first time for any company in a decade — including the most advanced chips in the world, Nvidia Blackwell chips.

It’s still early innings. This is an initiative that will play out over the course of decades. But where we are in the game, it’s looking pretty good.

Todd: When the CHIPS Act was passed in 2022, the US produced exactly 0% of any leading-edge logic or memory chips — all of which are critical to AI, and to all GPUs that Nvidia manufactures, not to mention iPhones and other infrastructure. We are on target to produce — by the end of this decade — 20% of all leading-edge logic chips in the world, and up to 10% of all leading-edge memory chips by some time in the mid-2030s.

Those massive investments draw very significant investments up and down the supply chain, which creates ecosystems that become self-sustaining, which creates dynamism in the economy. We are not there, but all the announced deals, and what’s being done on the ground now, should get us there in that timeframe.

The bill passes late in 2022; you’re hired in the months following. By 2023, what are you doing day to day?

Mike: The Secretary said, on my first day of work — at the White House podium — we were going to put out our Notice of Funding Opportunity (NOFO) in six months. A NOFO is how you describe what the application’s going to look like, how you’re going to evaluate it, how you’re going to get your deals done — “Here’s what we want companies to do.”

So those early days are figuring out, “How is this thing going to work?” We put out our Vision for Success.

Mike: It’s the basic architecture that you then have to fill out. As you’re designing the big picture, you begin to think through the small picture. We were doing a lot of resource planning. Todd, you remember those early meetings with the Secretary?

Todd: “How many people do we need?”

Mike: “Are we going to be ready? When are the applications going to come in?” Trying to project that out. Once the NOFO was out there in February of 2023, we wanted to be ready for contact.

The funny thing was, it took companies longer to apply. There was this weird period over that summer of 2023 where we were waiting. There’s plenty to do in terms of standing up the program and hiring the team. But it wasn’t until late summer or fall of 2023 that we were actively evaluating and negotiating.

That’s because a company like TSMC is going to take its time figuring out how to angle for $7 billion in government incentives?

Todd: We control when we put our funding opportunity out and the rules of the road. We don’t control how fast someone responds. We set the process up so that these leading-edge companies could apply quicker than others, because we thought they were going to be ready to go. I think we got Intel’s last application in September 2023 — six months after we put out the NOFO. You can’t control these things.

Sara: It was a little bit of a blessing, because by the time we hit June, we were at 100 people and we were reasonably ready to rock and roll.

The development of the NOFO is the thing you negotiate with the interagency. That first six months — from August to January — there were five official employees, between 0–30 detailees, and negotiations with the interagency about what should be in and out.

Mike: Six months to write a NOFO means three months to write a NOFO and three months of interagency negotiation.

We’ve talked about the interagency a bunch over the last two months on Statecraft — with Judd Devermont, who was on the National Security Council (NSC), and Dean Ball, formerly of the Office of Science and Technology Policy (OSTP). Were you all in the room with folks from this panoply of agencies? What were they pushing you on?

Mike: The White House set up some governance around the interagency that was hugely helpful. There was a CHIPS Coordinator that operated at the deputies level: Ronnie Chatterji [who spoke to Statecraft a year ago] and then Ryan Harper. The Coordinator was reporting into all the relevant principals: the Chief of Staff, the National Economic Council, NSC, and OSTP. We had a node of coordination.

Still, the interagency is a lot. You write a NOFO and send it to the Office of Information and Regulatory Affairs. Your NOFO goes out, not only to every agency with a potential interest, but to every White House office that interacts with those agencies. For every subject area — national security, cyber, environmental — you’ll have multiple sources of input, which might not always align.

What you’re sending around to the interagency — is this a Word doc that everyone’s adding their comments on?

Sara: That’s right. Then you’re responsible for responding to everything.

Mike: NASA wanted us to make clear that we’d fund semiconductor investments both on Earth and in space. We were able to say, “The statute requires it to be in America, which is Earth.” The way to move quickly is to accept comments. But if you accept every comment, you lose the thread on what you’re trying to achieve.

Sara: We didn’t have to write a NOFO. The NOFO required us to go through the interagency process in that more fulsome and traditional way. It was a choice that offered us structure, and protection from critiques that we were doing it in isolation. But it added time to the work.

Todd: Early on, we faced a bunch of criticism about “Everything-Bagel Liberalism.” Part of that is trying to find the balance between moving quickly and resolving interagency issues — and getting everything perfect. We had designed this in a way that allowed us to have a much more holistic approach [to evaluating applications]. We felt we could accept some of the comments from a bunch of the interagencies, move fast, and still have some flexibility as we made decisions.

For the first six months, it was all about getting ourselves up and running, and putting rules of the road out for those who were going to apply. The next six months was about starting to engage out in the world with potential applicants, and pushing them on what we would like to see, until we finally get those applications in, and start to evaluate them.

Phase three is the Preliminary Memorandum of Terms (PMT) stage. “Now we’ve got all these applications, how do we figure out how to enter into agreements on what these deals will look like, on a preliminary basis?” That’s a massive puzzle that you’re trying to put together, and use your $39 billion in an effective way. That was another six or so months.

The final stage is, once you have these PMTs, getting to legally binding agreements. That was a very tough process that got us to, at the end of the Biden administration, this $34 billion of deals.

You brought up the “Everything-Bagel Liberalism” critique…

Mike: Todd brought it up, just to be clear. Thanks, Todd.

This was a phrase that Ezra Klein originally used — the idea that the Biden administration wasn’t picking specific targets and then building programs around them. The CHIPS program was one of his examples. A component of CHIPS that came in for a lot of heat was that companies with large awards had to have a plan for childcare in their site. That’s one example of a broader critique of the Biden administration, and it’s a critique I’m sympathetic to.

I know all three of you have thoughts about that.

Mike: The core thing that the criticism missed was that we had set up an evaluation and negotiation process that was iterative and holistic. Our North Star was economic and national security. We were very clear about what that meant in the Vision for Success. For childcare, the statute required that we asked for a workforce plan.

Congress made you do that.

Mike: I think that’s fully appropriate. Workforce was a hugely important priority. We wanted to see childcare as part of that workforce plan. In terms of programmatic impact, I think we ended up getting some positive commitments from applicants. But it was not a sticking point.

It didn’t slow down the process?

Mike: Definitely not childcare. I spent hundreds of hours negotiating with these companies. I probably spent 30 minutes talking about childcare.

We thought that childcare, and workforce in general, was an important part of meeting our overall goals. Where we ran into problems is where there’s an institutional actor outside of the Department of Commerce that controls your fate and that may have different incentives. Those can be real choke points.

Todd: If you’re doing industrial policy like this, where the main thing is national security, that needs to be the priority. It is hard to stop other priorities from inundating you when there’s a lot of money around. We didn’t get it perfectly right. If we’re successful, CHIPS will drive jobs, but in my mind, it’s not a jobs program.

Micron, when this was all going about, was building a big childcare facility in Boise, because they recognized that in order to get people to come and work for them, they needed to provide childcare. If you go to Taiwan or Korea, all of these ecosystems have massive childcare facilities, because it is all about talent. If you talk to the companies about the biggest risk to this program, workforce is going to be in the top one or two.

I want to hear where you made hard calls in the interagency to keep stuff off the menu.

Mike: We had a fair amount of friction with applicants on requirements related to national security: cybersecurity, operational security, and the use of Chinese equipment. We were making some discretionary calls around what expectations we wanted to set, and they required a fair amount of negotiation. This is all on top of the national security requirements of the CHIPS Act, which constrained, for example, investment in manufacturing facilities in China. There were real trade-offs and healthy tension within our teams.

Sara: Something that we rejected out of the interagency process: we were pushed pretty hard to say what recipients were going to have to report to us upfront, as is normally required in a NOFO. We were able to say, “We are not ready for that. We have so much more to figure out here.” It ended up taking us two years to figure out exactly what that reporting was going to look like — how to make it not overly burdensome, and give us the data that we needed.

Mike: I remember you calling me and explaining, “With the Paperwork Reduction Act (PRA), we’re going to have to publish our reporting requirements for companies in the next few weeks.” My reaction was, “No, we’re not,” because we knew these were commercial things: we were going to have to negotiate it. Whatever we were going to have to do — emergency PRA exemptions, whatever pain we were going to have to take with the Office of Management and Budget — was necessary.

Why is that such a big deal?

Sara: Designing what people are going to have to report is always a big deal in government. There’s an inherent tension between getting everything the program might usefully use and anything that a recipient might want to do — it’s a huge burden. As somebody who’s been a recipient of federal dollars, all that reporting stuff might make sense to the federal agency. It makes zero sense to the people who have to do the reporting.

PRA makes it so that you have to pretend you can understand the trade-offs between the burden and the benefit in advance. [Statecraft has covered the Paperwork Reduction Act’s failings.] In this context, it’s even more complicated, because there are different deals, actors, and operating models on their side. Not to mention that we have to build the reporting system to ingest all this information.

Todd: This is not an interagency thing, but there were lots of internal debates about stock buybacks. These were a big thing for certain parties on the left.

People opposed companies buying back their own stock if they received CHIPS funds.

Todd: Obviously none of our money could be used to do stock buybacks. That was in the statute. The question was, “Can companies do stock buybacks at all?” You’re trying to incent the private sector. 5%-15% of the total money was coming from our office. The majority was being put in by these companies — who have to go out and talk to investors every single day. What you’re focused on is getting these companies to make significant investments in the US. But if there’s money after that, allowing them to give some back to shareholders is not crazy. Trying to focus on, “What are we trying to do?” and then allow companies to engage with the investors that they need — it was very complicated.

Let’s talk about Davis-Bacon, a law that requires federally-funded public works contractors to pay construction labor the local prevailing wage. This posed some issues?

Todd: Davis-Bacon was passed in 1931 — the Herbert Hoover administration — in an effort to create local construction worker markets. Many things about it are a bit anachronistic, but we all agree with the intent of trying to pay local construction workers the prevailing wage. The challenge is it had never been applied to something of this magnitude. There were tens of thousands of workers on a site like TSMC’s in Arizona. They come in and out, because they’re different types of contractors.

We weren’t paying for 100% — the companies were paying for a significant amount. We were trying to get people to start their work before they got their award, so we didn’t lose time. That created a retroactive issue: “What happens to all the work that was done before you got an award?” That had never been dealt with at the federal level for Davis-Bacon. That was the challenge that we, the Labor Department, and others were trying to deal with. It’s a whole administrative burden: how do you go back in time, for tens of thousands of workers, with thousands of subcontractors, and figure that out?

Mike: The industry got that it was part of the deal — it was in the statute and manageable. For fabs under construction, our strategy was to say to companies, “You’re building Fab One. We want you to build three fabs, so we’ll fund a little bit of Fab One, and then use that as negotiating leverage to get commitments for Fabs Two and Three.” But taking money on that first fab triggered this retroactivity issue. You need to look back at every contractor who’s done work on that fab and figure out if they were paid Davis-Bacon wages. That requires tracking down potentially 20,000 employees who are no longer on the site. That administrative challenge was very difficult to work through with applicants.

I’m seeing a through line with laws like the National Environmental Policy Act (NEPA), which we’ve talked a lot about on Statecraft, and Davis-Bacon. In these cases, the concrete requirement of the law — paying a little bit extra in wages — is not as burdensome as the process is. Is that right?

Mike: I think there’s probably some similarities. Both NEPA and Davis-Bacon are governed outside of the implementing agency. Davis-Bacon by the Department of Labor, NEPA ultimately ends up in a court, and a judge determines your fate, which is a terrifying thing if you’re trying to execute industrial policy.

In the Davis-Bacon case, you have a high volume of regulations and administrative case law built up over time. It’s mostly in contexts very different from semiconductor manufacturing. It’s a question of, reckoning with the existing body of administrative law in this new context.

Similarly, you’ve seen an increase over time in what NEPA expects, and that’s driven by litigation. Judges can ask for more every time. There’s case law that triggers an expectation that the implementing agency is going to do more for NEPA, and that accumulates.

Back to the CHIPS timeline. We’re in late 2023–spring 2024. You’re starting to try and get these deals done. You’re negotiating with some of the biggest companies on Earth. What’s your strategy?

Todd: PTSD is coming back. I think the most complicated period was trying to get these Preliminary Memorandum of Terms (PMTs). We put preliminary offers on the table starting around Thanksgiving of 2023. Then it took time — between November and April — to negotiate those. We had $70 billion of requests from just the leading-edge firms that were going to get some subset of the $39 billion. The expectations were way too high, so we had a lot of negotiating. You have to figure out how much you’re giving to each of them, you’re negotiating with all of them at the same time, and trying to get that puzzle to come into place.

Leading-edge companies between them asked for something like $75 billion. You ended up giving them ~$25 billion. In the process of negotiation, you cut them down two-thirds?

Todd: That’s correct. And everybody focuses on the big numbers, but there’s also a whole series of, “When are we going to get that money? On what milestones? How much comes early versus later?” They’re the most valuable and sophisticated companies in the world for a reason.

These companies are hard negotiators?

Todd: They’re tough negotiators — and there’s a reason for the CHIPS Act. TSMC has a credible option to do more fabs in Taiwan. Micron has fabs in Singapore, Taiwan, and Japan. They’re making a real choice: “What is the amount of money that will make it positive for us to put all this capital?”

Sara: By the time we got those PMTs on the table, everyone was exhausted. We had done so much process-design work — every single thing from accepting the applications to how the Investment Committee was going to work. If you asked us a year later, I don’t think we realized how much more intense it could have gotten.

You’re sprinting to stand up this thing, and at the end of the sprint, you start negotiating with some of the biggest sharks in the world.

Mike: Some of them were sharks. Maybe not all.

Sara: When you are standing up a program from scratch, you have to design every single process you run. That is a gift: you get to choose and learn from all your prior experiences. And it’s real hard. The minute the process you’re focused on is stable, you’re onto the next thing. We were designing, testing, implementing, and then just repping out processes, from January 2023 through January 2025. We did not stop.

Mike: Our first PMTs were two small deals. We announced BAE, a small defense manufacturer, for $35 million, and Microchip was $162 million. After we announced BAE, it felt like this Herculean thing. We had an all-staff meeting right after, and I said, “Congratulations. We just need to do 1,400 more, and we’ll be done with the program.”

The core governance of our investment process was the Investment Committee. It had to evaluate the economic and national security benefits, sign off on the merit review — but another important function was figuring out the right size of the award. We had created a framework for sizing awards based on the rate of return of the project. The basic idea was, “We need to make this project make sense for you economically — to make your hurdle rate. How much of our money is going to be necessary?” That is as much art as science.

For those early deals, we were spending hours in the Investment Committee, trying to figure out how to make this internal rate of return analysis work. A lot of people on the team were terrified. People on Todd’s team who we were torturing with all our questions were like, “If you’re asking about this for the small deals, how are we going to do the big deals?” But every time we went through something new, we had to build the muscle. If it wasn’t working well, we had to go back, reiterate, and do new process design — to create as much efficiency as possible, while maintaining the standard we had set for ourselves in terms of analytical rigor.

Todd: This was modeled as a world-class investment firm, where the returns were to national and economic security. Me, with an investment background; others on the Investment Committee that come from the semiconductor industry and had deep technical expertise; Mike with his background in government — trying to get everybody around a table to understand what a good investment looked like. When you’re at an organization, you don’t even recognize the norms and processes that are in place. We were putting all that in place.

You came from a professional environment where these are all fundamental norms. You had to rebuild this for a bunch of folks who have no experience in the boardroom?

Todd: The core investment team all had experience there. The goal of any investment decision-making is, “How do you get people with diverse expertise around a table, where everybody’s incented to voice their views, and make good decisions?” That’s what we were trying to embed. We’d hired many people who were working at the world’s greatest investment firms — Blackstone, Goldman Sachs, VC firms. We had a whole group that had semiconductor experience, that were chief technology officers at GlobalFoundries, worked at Intel, or in the industry for 30 years. Mixing that with the relevant expertise in national security, workforce, and environmental was the challenge and, ultimately, the magic.

BAE and Microchip are the two “small” deals you land. Then TSMC is the first big one?

Mike: I think we announced Intel first, TSMC shortly after, and then Samsung and Micron. Those are our big four.

Mike, you and I were talking recently about you flying to Taiwan for the negotiations with TSMC.

Mike: We had signed our preliminary term sheets with the big four. We had negotiations ongoing with a set of smaller, but strategically important deals — companies like Amkor, SK Hynix, Entegris, Texas Instruments, GlobalFoundries, Legacy.

To get these term sheets to a final award, we knew we’d have to negotiate a lot of the details around policy commitments, like workforce and environmental. There was this general sense that the lawyers on both sides were going to figure out the detailed terms. But when we came up with our draft award documents and sent them out to market, the initial reaction was very negative. Not on the policy stuff, but on, “What is the core legal relationship between the United States government and these companies?”

What were these companies upset about?

Mike: It was conditions precedent, representations, warranties, and covenants…

Todd: It’s all about who bears the risk: the US government, or the company?

Mike: “When can we stop funding? When can we call funding back? How much discretion does the government have? When can the government or the company terminate the agreement?” The default in federal programs is, “Give the government a huge amount of discretion.” But that is not something that these companies were used to seeing when dealing with Singapore or Japan. It became something we had to manage very aggressively.

They would say, “We’re nervous. This agreement is going to last for 10-plus years. Who’s going to be at the CHIPS Program Office in 10 years?” They were making decisions on billions of dollars of investment, based on how much money they were getting. They wanted certainty that, “If we hold up our end of the bargain, we’re going to get that money.” It was a very real and fair discussion.

You hadn’t expected to have to do that — you thought this is primarily the lawyers’ job?

Mike: We had hoped that it could be worked out in the legal details. Very quickly, Todd and I had to become personally involved in understanding the structure of the agreement, where the pain points were, and navigating towards a solution. We spent days in a conference room in TSMC headquarters working through these legal details. I remember one of our lawyers explaining to them what the Fly America Act is. I’m sitting there thinking, “What is the Fly America Act?” It’s a cross-cutting statute that says, “If you receive federal dollars, you can’t support travel that isn’t on US airlines.” We’re explaining that to them, and they’re saying, “How can this work?”

You’re kidding? They have to document that when they fly to the US, they’re not flying on foreign flags?

Mike: For the Fly America Act, it ended up being, “If you use our dollars to do it.” It’s just an example of what you’re wading through within our legal construct. We did a week in Taiwan, then the TSMC team came and did a week in DC. Everyone was tired and wanted to be done, and we were grinding through the details. To keep morale up, we worked with Ryan Harper to do bowling at the White House bowling alley with the TSMC group. It was great for the deal and our relationship with them.

But I remember asking their key lawyer, “How does our contract compare to Japan’s contract?” He goes, “We don’t have a contract with Japan. We just submit evidence of our investment, and they give us the money.” That probably undersells the complexity of Japan’s system. But it was such a stark indication of the difference between our system of government and law, and other countries.

Todd: In June 2024, nobody on the team thought we were going to get all these deals done, because it felt like the system was clogged up. Mike and I felt we had to get involved. If we could land TSMC, and say, “The most important company in the semiconductor industry has agreed to this,” we could then move much more rapidly with others. That ended up being true, but it took longer.

Take Intel, for example. We sent the first [draft] PMT on Thanksgiving of 2023. I flew out on the Monday after Thanksgiving to Santa Clara to meet with their CEO to talk about it. We announced the Intel PMT on March 20th — so four months to finalize, negotiate, and get all that in. We thought that once you had these PMTs in place, the next stage would be easier. Lots of things happened post-PMT, including Intel’s August results announcements, where their stock went down by a significant amount — they ultimately fired their CEO. This space between, “You have an announced agreement and everybody’s excited because we think we’re close to the end,” and getting a final agreement in place, was very complicated.

Mike: It required a huge amount of governance within the team, because we had to figure out how to make quick decisions on what the companies were asking for. Sara stood up internal governance committees so that she could resolve issues, or we could escalate them and design together. You might be spending eight hours with a company in an office. You spend three hours negotiating, then maybe you break up for an hour and quickly call the general counsel, or come to a decision internally, and come back in the room and say, “We’re good to go with X. Can you deal with that?” They would often say, “We’re still waiting for legal to get back to us,” or, “Our head of finance is on vacation.” We were feeling so much urgency, and we had created a pretty nimble structure — it was the bureaucracy of the companies that was sometimes holding us up.

What’s the structure that let you move faster?

Sara: We had office hours, and a steering committee to litigate between the teams: what we would be looking for in due diligence, what our terms in that first iteration of the term sheet would be. Once you create some structure and do a couple of reps, you have precedent for some decisions.

Todd: Two times a week we would get together and go through our priorities: where we were, what needed to happen. You need to have a willingness to make decisions quickly, then a structure that says, “What do we have to do this week?” Then later in the week, “What haven’t we done?” Creating the battle rhythm in any organization to make decisions quickly is critical.

Mike: That started with the Secretary. When we got to this phase, it was every Monday morning with her. We would print out our tracker, deal by deal, and talk about where we are.

The other thing is we had a hugely constructive and regular dialogue with the General Counsel of the Commerce Department. She always knew exactly where we were in every deal, what the pain points were. There would be times where we would have a company in the building. If we reached an impasse, we’d call her up and say, “We need you down here at noon. Can you make it work?” She’d come in and negotiate. Maybe the Secretary would stop by, not to negotiate, but to give a little pep talk.

What I’m hearing is that some of the things that were reported as roadblocks, like childcare requirements, weren’t. What slowed you down was that figuring out complicated financial deals takes time. In practice, where did the slowdown happen?

Todd: In the PMT stage, the commercial deal — “How much? Over what time period? What are the milestones?” — took by far the longest period of time. The next [source of delay] is the contract. That’s not specific, like Fly America — it’s typical contract negotiation in almost any large private-sector deal. You’re always trying to find, “Who’s got the risk?” and negotiate that. The government is used to having all the power. But these are big organizations that are trying to find a more reasonable — “Where are you on the fairway?”

Most of the other things that are talked about — workforce or childcare — did not take a lot. For some of the companies, there were specific national security things that we cared about. Those were important, but some took a lot of time. Then there’s the classic things like NEPA and Davis-Bacon, that took a bunch of time, discussion, and our own understanding. But those were one-offs dependent on individual company situations.

Sara: To the extent that there was a niche thing that was holding anything up — we wouldn’t have let it. We would have reorganized ourselves in service of the highest-priority things, if something in the panoply of requirements had threatened to derail the awards.

Mike: I think that gets to the core point, which is: “Where does discretion lie? Are there other institutional actors who can hold you up?” For Davis-Bacon, we did have some real challenges. NEPA, we ended up with a statutory carve-out for most of our projects. We had already posted public Environmental Assessments. There were some clear signals that litigation might be pending, and Congress intervened.

Todd: While it ended up not being that critical, because of Congress intervening, we knew that NEPA was going to be a big challenge. We set ourselves up to be able to serve that. We had an incredible environmental team that was very proactive, we asked for information, and worked with these teams early. NEPA ended up being legislated away — but even if they hadn’t, the way we got ahead of it — there’s probably some lessons there.

Sara: I think that is the core lesson. We can do a lot of complaining about how hard it is. Maybe Congress is going to come to the rescue — and also you’ve got to just get ready to do the work. In a lot of cases, we did a lot of extra work that maybe we didn’t need to, because it was a better use of our time to get to grinding than it would’ve been to fight the thing that was frustrating. That was certainly the case with NEPA. We would have done the work, and done it as well as we possibly could have.

Todd: I had no idea what Davis-Bacon was, and we didn’t realize that that was going to be an issue. We would’ve set ourselves up differently and tried to get ahead of it — we ended up being behind the ball there.

The way that we set up the decision-making process — we keep saying it was holistic, and we had a lot of flexibility. Most grant programs have this scoring mechanism — three people score it, average it out, and let the chips fall where they may. We didn’t have that. We had a much more holistic process, where we were clear that we had these six categories. The first category, national and economic security, was going to get the most importance in any decision-making. But there were no formal scores or anything like that.

While we had a lot of things in our application that said, “We need to see your workforce plan and your childcare plan,” that didn’t mean that there had to be something specific on that. We could make some decisions on, “This is so important to national security that these issues don’t matter as much.” Our judgment, to the Everything-Bagel point, was, we are comfortable letting people give us their childcare, environmental, and workforce plans, because we’re not saying, “And you must do this.” We’re saying, “We’re going to take that into consideration in a holistic way and make determinations.” We had the flexibility to decide what was in the final award documentation.

If one of these big companies that you felt, for national security reasons, we had to strike a deal with, hadn’t made the cut on this metric, you had ways to make sure you could still find a grant?

Mike: They needed to meet a minimum threshold for each of our six categories. But beyond that, we had meaningful discretion. That’s something I thought about while we were designing the process, because if we had done the default — six points for this, eight points for that — the comfort in that is, by getting rid of administrative discretion, you are getting rid of oversight risk.

You’re being fair.

Mike: We’re being exactly fair, we’re doing what we said. Except I had this inverse intuition, which is, if TSMC had ended up not meeting a threshold, for some reason that we didn’t foresee, because this was so complicated and designing evaluations up front is so hard — figuring out how to change the process so that it was above the threshold — that’s not a comfortable position to be in. I ended up with a good relationship with the lead lawyer on this — we would say, “We need to design this process based on how we’re actually going to do it.”

Todd: Typically, there’s not a lot of proactive engagement with the applicants. Most grant programs are set up where — put in your application, maybe we’ll set up one or two conversations — so that everybody’s treated the same way. We wanted to shape what was coming in. We didn’t want to just do the things that the companies wanted to do. We wanted to push them to do different things. We hired a team of people that could engage with these companies in a sophisticated way. There’s a little bit of risk in it, but we could regularly engage, push back, say, “You should do more here, because that would be more powerful from a national security perspective.” If you want to make sure you have zero oversight risk, you would have gone a different direction, but you might also not get as much done.

I think most people think the Inspector General roots out waste, fraud, and abuse; congressional oversight stops you guys from cheating — Todd is rubbing his eyes and Sara is shaking her head. You have this visceral reaction to that claim. [Sara discussed her experience with the IG at length here.]

Sara: I am fully supportive of there being inspectors general (IGs), the Government Accountability Office, and congressional oversight. It comes down to the execution. I have lived through experiences where they get it wrong — it’s so easy to get it wrong. In particular, when we’re conditioned to say, “The way the government’s supposed to operate is fully-documented, fully transparent, fully fair in everything it does.” It means that you have to do all of the things that we had the luxury of being thoughtful about at the outset: write everything down, then do it exactly that way. If you don’t, there’s a million threads to pull on. IGs get congressional appropriation, and they have to deliver work.

They have an incentive to have a big headline. One of the ways in which this has all gotten unhelpful, in the typical government experience, is we have so much fear of oversight. Having a constructive relationship with your overseers is an important thing that lots of programs don’t ever experience or try.

We tried to do that, and it makes all the difference in the world, because then you at least start to level the playing field in terms of the information-asymmetry problem. They don’t know all the context for the program you’re operating. We tried to anticipate, “What are all the places where, in our process design, it’s going to be important that we write down why we made this choice?” We erred on the side of more documentation than we needed.

Mike: There are two pathologies that emerge because of concerns about oversight risk. One is that you design a process to be immune from criticism. Our mantra was, “The biggest risk is that we don’t get this done.” If we’ve successfully implemented the program, and a year and a half from now there’s a negative IG report on the process, I can live with that. That doesn’t mean we’re not going to engage proactively and have documentation, but that is a trade-off I’m willing to take if, a few years from now, fabs are being built and chips are being produced.

The second is a paralysis around risk of bad outcomes. You do one deal that goes poorly, and that ends up driving a narrative about the program, headline risk, and congressional oversight. You want to have all the analytical rigor deal by deal. But we thought as much about risk of omission as risk of commission, and we thought about risk at the portfolio level, not just the deal level.

You had this portfolio of companies, and you weren’t expecting that every single investment would be a massive success?

Todd: Inside government, you always hear about “Solyndra risk” — something goes really sideways. A knee-jerk reaction is, “We don’t want another Solyndra.” I think that is a very wrong way to frame things. If we’re eliminating risk from the portfolio — we’re not doing things that have no risk of failure — then we’re not doing anything that the private market won’t do themselves. We need to take educated bets, and do good underwriting and analysis.

We had this robust debate with Secretary Raimondo about, “We need to make sure the world understands that we’re trying to incent something, and not everything’s going to be successful. But it’s okay if some of our investments don’t go as well as others if we’re accomplishing the ultimate goal.”

In leading-edge logic, we gave awards to all three — Samsung, Intel, and TSMC. What we said up front is, “We need at least two ecosystems in this country to be successful at leading-edge.” The market’s going to sort that out, and hopefully all three will be, but at least we know we’re on target for getting that 20%.

Sara, will you tell me about some of the challenges with the IGs?

Sara: We were wildly successful at hiring very quickly. Our average time to hire was something like 67 days at the end of 2023, after the big surge that we did. The benchmark for the government is 80 days, and I don’t know that agencies have ever actually achieved that. It’s more like 101 days, the last time I looked at the Office of Personnel Management (OPM) Dashboard. We hired more people than we expected to be able to in that first nine-month sprint. The whole time I’m there, I am anticipating when the first call from the IG is going to come. We get the first one, and it’s about hiring, which makes sense.

Hiring’s always a big risk. You could get a bunch of money from Congress, and if you don’t have the people in place to deliver the program, you’re going to be behind. We felt pretty confident going into that. We handed over all the data and answered all their questions. We get the draft report back, and the headline is, “CHIPS succeeded in hiring all the people that they wanted to hire, they exceeded their hiring goals, but they did not develop a comprehensive workforce plan.”

What’s a comprehensive workforce plan?

Sara: I don’t even know — I’ve been in government for a long time. This is a recommendation from OPM that you do this multi-stage planning document, which makes sense in some abstract way.

What did they want you to have developed?

Sara: Literally the number of positions — in government HR language, “What are the functions and capabilities that you need? Why? How many people in each thing and over what time?” I’m sure there’s all kinds of stuff about estimating attrition and whatever to come up with some very formulaic answer for what you should then go hire for.

I have actually never conducted one of these things. I think usually they are for an entire agency. They were applying this standard to us. I’m sure there’s also history with the work that they had already done to look at the National Institute of Standards and Technology and Commerce in the past. They’re bringing some of that baggage to it. But that is the headline, and it takes up two thirds more real estate than the fact that we succeeded.

That you hired ahead of schedule?

Sara: Absolutely absurd. We had hired ahead of schedule, faster than the benchmark, and had a highly efficient process.

Mike: Not to mention, we’d built an unbelievable team. They didn’t even get into that. The mindset that one has, to provide that feedback — there’s such a delta between that process bureaucracy and what it took to build the team.

There were some notions about what the org would look like. I thought I needed to think about it. The Secretary said, “You have two weeks.” I met with her two weeks later. I laid out a vision for what the organization would look like. I’d say, “We’re going to have an investment team, a strategy team, ops, risk, legal, external affairs. We need to find all-star people to lead these teams.” Todd was going to lead investments, Sara was going to lead ops and be chief of staff. “Find those people.”

When Todd took over investments, she was like, “Todd, what do you need?” It was that level of urgency, reporting weekly to the Secretary on how the team was being built. The Secretary designated her most senior person within the Secretary’s office at CHIPS to work full-time on supporting recruitment.

Todd: Her message was not, “I want to hold you accountable.” It was, “Tell me where I can be helpful.”

Mike: She’s very charming. You get her in the room with someone, and she would always close.

Sara: The IG asked us, “If you didn’t workforce plan, how did you decide? Where are your planning documents?” I said, “We got the leaders of each of the organizations in, and we said, ‘Run and find the best people to do the job.’” You could trust that the humans that you put in the seat had the right experience to design this. But that almost struck them as foreign. They criticized it in the report.

I do want to hear more about how you did hire, because you hired a lot of top talent with the capability to execute these deals — talent that usually you cannot get into the federal government.

Todd: We can attract these people to government. I just don’t think we have tried that much.

Sara: We had ample administrative funding and no real Full-Time Equivalent cap — we could hire the number of people that we needed. In the CHIPS Act itself, we had an authority to hire up to 25 people with higher salaries — more than Senior Executive Service roles. It’s nothing like people are making on Wall Street, but it is significantly more than most civil servants make. We also asked for Direct Hire and Excepted Service authority from OPM. They give you an easier, faster process to get folks in the door, and one that gives you more discretion in making your selections.

One thing that is worth repeating — Statecraft recently discussed it with Judge Glock — is that most people in the federal government can’t identify a talented person and say, “We’d like to hire you.”

Sara: That’s correct. In my 16 years in government, I don’t think I have ever had the benefit of having an Excepted Service or Direct Hire slot. It is an enormous advantage. It has a lot to do with our average time to hire, because we relied on those slots, in the first instance, for nearly all of our positions.

On recruiting private-sector folks to government — it is a learning curve for everybody. But having the diversity — people who are not typically in government and people who’ve spent some time there — was pretty magical, and hugely important to our success and credibility.

Todd: We needed different types of people. We were going to have these small deal teams, similar to consulting or investment firms, that would be the Intel or the TSMC team. That drove the kind of people you needed. We needed a senior person that had the experience, the gray hair, and could stand side-by-side with the CEO of Intel. And we needed some mid-level and younger people who were strong analytically, strategically, and financially.

There are a ton of people out there who have a strong desire to work in government and give back — and believe in the mission. We did get a number of close-to or at-retirement people — 30 years at Goldman Sachs, or Lynelle, who worked in the industry for decades and was retired. We got a number of them to come back out of retirement, which was amazing. I think the most impressive thing was to get these young and mid-level people — who were climbing the ladder in the private sector and making real money — to take a risk and come into government. They knew they liked the mission, and they wanted to do something where they could use their skills.

If you now look, the vast majority of those people are back in the private sector doing cool things. Their career trajectories have taken off. They all are going to look back on this period of time as unique and transformational — and they’re back. If more people would feel like, “I could go do something in government for a handful of years and then go back into the private sector,” that would create more dynamism.

Sara: It is also self-reinforcing. Todd, you and Sara being recruiters, and then your senior people being recruiters — that sends a signal to these younger people. “You have real serious people who have experience that is relevant to me.” I imagine that that has to have played a real role.

Mike: Todd and I went on Odd Lots in April ‘23, and we probably had five or ten people who listened to that, thought, “That sounds cool,” and ended up working on our team.

It would be a mistake to say, “We brought in this private-sector talent, and that was the magic.” We also attracted unbelievable government talent. One of the cool things was watching the mutual respect emerge, as people learned to work together. Government lawyers talking to Taiwan at 11pm every night for a couple of weeks to get a deal done — and then being up in the morning to turn the docs. It was incredibly arduous. It is extraordinary what people will do if they feel connected to the mission.

One of the hard things about procedural barriers — whether that’s an internal process design or an external barrier — is you feel it immediately in the team’s morale, because the teams are saying, “We’re trying to do something important. Can’t you solve this for me?”

Over the coming year, we’re discussing many of the lessons you learned in Factory Settings. I think this is a good lesson to end on. You spent an inordinate amount of time actively hiring, and trying to get senior politicals engaged in the hiring. It’s not just that you had on-paper hiring authority from Congress to make these discretionary calls — you have to go and use it.

Todd: Mike used to say, “We need to view the bureaucracy as our dance partner.” It’s not that we need legislation, or everything needs to fundamentally change. We need to work with some of the rules that are in place as well. There’s things that can happen tomorrow without any changes in government. That’s a lot of what we did. We figured out ways to take days out of onboarding people. There are a million different things, but they all add up.

The reason we’re doing Factory Settings — we were having dinner at Mike’s house one night, and he asked me whether I thought we were going to get this done. My comment was, “Yes, but nothing about it seems repeatable or sustainable.” Our goal is to try to impart some of those lessons so that it can be repeatable — it’s not just a bunch of people doing 24/7 for two and a half years to break through everything. Rather, there’s some structure around it that is repeatable and transformational over time.

• Sufficient programmatic resources (between the 25% investment tax credit and our $39 billion in funding)

• Substantial administrative budgets to build our team and procure contractor support

• Special hiring authorities that let us attract great talent and scale rapidly.

• “Other transaction authority” that let us sidestep grant law strictures and negotiate with industry.

• Nimble, dynamic industry engagement models.

• Substantial statutory discretion to design our process and make funding decisions that advanced the national interest.

Today’s post focuses on an area where our “factory settings” initially didn’t diverge from federal baseline requirements: the National Environmental Policy Act, or NEPA.

NEPA created significant risks of delay and litigation for the program. We managed those risks through early engagement with applicants, careful process design, and a highly capable environmental team, but the threat of disruption remained. Ultimately, Congress provided relief by narrowing NEPA’s application to the program, reducing — but not eliminating — the risk that environmental review would derail projects, or the program as a whole.

What is NEPA?

My first call from Secretary Gina Raimondo about NEPA came a few weeks into my role, while I was cobbling together dinner after putting my toddler to bed. Her inquiries usually followed a formula — raising massive topics that reflected our implementation challenge but for which we had no plan. This call had particular urgency: we needed a plan for NEPA.

The urgency around NEPA stemmed from the dual risks of delay and litigation, each of which represented a source of potential program-level failure. To understand why, it’s helpful to know what NEPA is — and what it isn’t.

Semiconductor fabs, like most industrial manufacturing operations, have substantial environmental footprints. They use land, consume massive amounts of power and water (though they recycle most of the water), emit fluorinated gases and other pollutants, use hazardous materials, and require disposal of hazardous waste. Starting in the 1970s, we enacted a broad regime of environmental laws — laws like the Clean Air Act, Clean Water Act, Endangered Species Act, and the Resource Conservation and Recovery Act. Over time, an extraordinarily complex regulatory regime has grown from these statutes, comprising hundreds of implementing regulations. Layered on top of these laws, state and local governments also have substantial permitting regimes: for example, TSMC’s Arizona fab required more than 600 local permits, compared to around 60 in Taiwan.

NEPA, by contrast, is not a permitting statute: it imposes no substantive requirements, other than those that might emerge through litigation. Instead, it adds a procedural requirement for the government to assess the environmental impacts of so-called “major federal actions.” Early in my tenure, I probed whether our CHIPS investments had to qualify as “major federal actions.” But accumulated regulation and case law made it clear that each of our projects would need to satisfy NEPA before federal support could flow. I also probed for other areas of flexibility — might there be a NEPA exemption for projects critical to national security? — but found none.

NEPA’s requirements hinge on whether an agency’s proposed major federal action will have a “significant impact” on the environment. The baseline requirement is for an agency to produce an “environmental assessment” (EA), which typically take 9-12 months to complete (and often take longer than that).1 An EA can result in a “Finding of No Significant Impact” (FONSI), or it can condition a FONSI on environmental mitigations (a “mitigated” FONSI). A FONSI or mitigated FONSI allows the agency to proceed (subject to litigation).2

If a major federal action is deemed to have a “significant impact,” the agency must issue an “environmental impact statement” (EIS), evaluating the proposed project’s potential environmental, social, and economic effects. On average, an EIS runs 1,703 pages (including appendices)3 and requires an average of 4.4 years of work.4 It must consider “reasonable alternatives” to the agency’s proposed action, including a “no action” alternative, and thoroughly analyze the action’s “reasonably foreseeable” direct and indirect impacts. The agency releases a draft EIS for public comment, then produces a final EIS demonstrating it considered all comments — a major driver of the multi-year timeline. Lawsuits then add more time: the median lawsuit challenging an EIS lasts 1.6 years, but 7% of lawsuits last more than 6 years.

NEPA’s force derives from the fact that anyone with standing can sue the government and challenge its findings. In legal terms, the Administrative Procedure Act gives citizens the right to sue agencies and seek judicial review of any “arbitrary and capricious” decision-making — including any EA or EIS that they can persuade a judge reached incomplete or flawed conclusions in its environmental analysis. For example, an EA lawsuit might contest the finding of no significance and argue that the agency should have to prepare an EIS. An EIS lawsuit might allege insufficient analysis, inadequate consideration of alternatives, failure to assess cumulative impacts, or deficient response to public comments. Worse, the general statute of limitations under the APA is six years from the agency decision — meaning the litigation risk on a NEPA review lives with the agency and project sponsor well after the project is approved and can potentially disrupt implementation at any time.

For CHIPS, the immediate source of potential delay was self-evident: we needed to accelerate the typical timelines for writing EAs and EISs to account for the commercial realities of one of the world’s most dynamic and competitive industries. But cut any corners, and you increase the risk of losing a lawsuit — which could result in an injunction on either the CHIPS funding or the project itself. The looming threat of an injunction is exacerbated by the fact that NEPA itself imposes no objective, substantive requirements. The fate of the “major federal action” — and therefore the project it’s intended to support — depends on a judge’s assessment of the quality of your EA or EIS. As an implementer, there’s no objective way for you to know if you’ve hit the target.

Throughout the program, NEPA consumed significant leadership bandwidth: regular check-ins with my team, periodic principal-level meetings at the White House, and frequent engagement with applicants to manage the process. During one particularly intense stretch in the winter of 2023-24, while we were simultaneously negotiating our four major leading-edge deals, I analyzed my calendar and found I was spending nearly a quarter of my time on NEPA alone.

Subscribe now

NEPA in the context of startup industrial policy

While NEPA requires effective execution across government, we faced operational and commercial challenges specific to a startup program doing large-scale industrial policy:

Building the team

To oversee the NEPA process and other environmental issues, we hired Komie Jain, an executive with deep experience in environmental review and permitting across both industry and the federal government. Komie’s background was well-suited to a role that required forcefully representing the government while remaining sensitive to the commercial realities facing our applicants. Komie’s team had two units: one responsible for meeting our obligations under NEPA, and another charged with managing permitting issues — evaluating the permitting readiness of applications, and proactively engaging with applicants and federal, state, and local governments to resolve bottlenecks.

Across both units, we built a team of 12 FTEs quickly, using our special hiring authorities. For the NEPA unit, we generally hired people with prior experience of writing EAs and EISs at other federal agencies — NEPA is a bespoke process with its own internal language and logic, so this experience proved invaluable. We also secured contracting support that gave us access to up to 20 contractors as an additional surge resource that we deployed regularly.

The environmental unit was housed under the Strategy vertical, which comprised teams focused on strategic and policy priorities (economic security, national security, workforce development, and environmental review and permitting) across all deals. By contrast, our Investments vertical organized teams around specific deals.5 The environmental team was the largest in Strategy, reflecting the operational intensity required to fulfill our obligations under NEPA.

Industry engagement

A NEPA document is ultimately a government work product, and the government must defend it in court. However, the only practical way to complete several reviews simultaneously on tight timelines is to rely on applicants to do the underlying analysis and drafting. Not all agencies allow this — some insist on keeping the drafting in-house — but given our resource constraints and commercial timelines, we had no choice.

That starts with gathering information. We developed a substantial environmental questionnaire as part of our application to jumpstart the process. Some criticized it as overly burdensome, but its requirements were driven by the requirements of NEPA and the risk of losing a lawsuit.

Because we worried about NEPA driving delays, we wanted to start engaging with industry immediately. This meant beginning environmental review concurrently with negotiations over investment plans and incentive levels — usually while firms were still considering international alternatives. This created a somewhat awkward commercial dynamic that required our environmental team to engage firmly while recognizing applicants’ commercial constraints and international alternatives. Moreover, while we could lay the foundations for an efficient review process early in negotiations, we couldn’t actually begin drafting until we had reached a non-binding term sheet with the applicant. After all, the term sheet would define the scope of the project, which in turn would determine the environmental impacts.

To support the drafting of EAs and the associated underlying analysis, applicants needed to hire environmental consultants who specialize in developing NEPA documents. Then there was substantial back-and-forth between the government, government contractors, the applicant, the applicant’s consultants, and everyone’s lawyers to refine analysis and edit the document until both the government and the applicant were comfortable issuing it. The process’s efficiency depended on the strength of all those parties as well as the underlying facts. One deficient cog could shut down the whole machine until remedied.

The EIS challenge

For an EIS, the challenge intensifies by an order of magnitude. After detailed evaluation of each project, we ended up needing to write just one EIS: Micron’s project in upstate New York, which will build fabs to produce the leading-edge high-bandwidth memory chips at the center of the AI boom — all currently made in Korea, Taiwan, and Japan. Micron’s EIS was triggered specifically by the presence of wetlands and endangered bats on the project site, pulling in the US Army Corps of Engineers and the Fish and Wildlife Service. But it also required detailed chapters covering numerous other impact areas: land use, geology and soils, historic and cultural resources, emissions, solid and hazardous waste, human health and safety, utilities and supporting infrastructure, traffic, noise and vibration, visual effects and community character, community facilities, socioeconomic conditions, and environmental justice — each expanding the surface area of legal attack.

To prepare the EIS in collaboration with our team, Micron had to make an enormous investment into investigation, analysis, and drafting. Within CHIPS, we dedicated multiple FTEs and substantial contracting resources. We also had to coordinate between four federal agencies, nine state agencies, three county agencies, and four town boards.6 Within each entity were often both program offices and legal offices, and each chapter generally required separate legal review. I personally held weekly meetings with my White House counterpart (who helped coordinate the other federal agencies involved in the review process at senior levels) and Micron executives to manage the project given its significance to our program.

We signed our preliminary term sheet in May 2024 and set an aggressive schedule to complete the EIS before the spring of 2025 — Micron needed to clear trees before bats returned from winter migration for the nesting season. We ultimately missed the spring 2025 tree clearing window and reset our target for the winter of 2025.

Commerce just posted the final Micron EIS and record of decision. This may seem like a long time, and it is, but compared to the average time to completion of 4.4 years, finishing an EIS in 18 months is a huge accomplishment for the federal government, the state government, the county government, and the company. The final EIS ran 737 pages. Including its 18 technical appendices, the full document is approximately 27,000 pages.

Still, a project initially expected to produce chips by 2028 won’t produce any until 2030, and NEPA as a big part of the reason why. Meanwhile, the company is so supply-constrained selling into the AI boom that it just announced the extraordinary decision to stop selling chips for consumer goods like iPhones.

NEPA in a fiercely competitive industry

Semiconductor manufacturing is among the most fiercely competitive industries in the world, both for the companies within it and for the nations competing for manufacturing investment. NEPA creates specific disadvantages in this environment, including:

• Uncertainty for ongoing construction: Because we were funding projects already under construction, companies sought assurances that continuing to build wouldn’t adversely affect their NEPA evaluation. We were unable to offer clear assurances. This created uncertainty for companies making real-time construction decisions while negotiating with us.

• Commercial agility: Each leading-edge fab represented a $20+ billion investment on schedules tied to customer demand expectations. Product cycles are measured in years, and companies often need to adjust plans or make tough decisions based on the commercial environment. Project delays could divert demand to overseas fabs, or compromise commercial relationships that took years to build.

• International competition: We were also competing with countries like Japan and Taiwan to attract these companies. Both have mandatory environmental reviews, but they proceed with much faster timelines and without the same threat of litigation. Companies trying to maintain a competitive edge can see this difference clearly.

Subscribe now

Putting it all together

By the fall of 2024, two years after launching the office, we’d managed the NEPA process as well as we could have hoped. We’d issued draft EAs for public comment for TSMC’s Arizona project and Micron’s Idaho project, and were close to posting drafts for Intel and Samsung. We therefore felt reasonably confident that major projects wouldn’t be delayed by our drafting timelines. Micron’s New York EIS was the one exception, and even getting that on track to completion by the end of 2025 felt like a win. We’d also published a “programmatic EA,” a standardized environmental assessment allowed under NEPA to streamline future project-specific reviews for a broader category of projects. Ours intended to cover a range of smaller-scale fab modernizations and allowed us to greenlight many of our smaller investments.

I’d love to take credit, but I’m struck by how contingent that success was. A few factors were key:

• We hired a great team. As anyone who builds a big startup knows, you get some hires right and some hires wrong. Komie proved to be truly excellent: knowledgeable about the substance, highly effective in project management, and savvy in applicant interactions — commercially sensible yet forceful in explaining what we needed. A poor or even average hire would have set us back months.

• Second, we developed effective working relationships with applicants willing to invest heavily in the contracting support necessary for analysis.

• Third, we ended up with only one EIS — an outcome highly uncertain at the start. At one point Sara Meyers walked into my office and casually told me another high-profile project would be an EIS. She says she’ll never forget the look on my face. (It turned out to be a miscommunication.)

Even with our successes, NEPA’s ultimate impact on the program remained uncertain. Based on comments we received on posted draft EAs, our lawyers expected lawsuits — at that point, our fate would rest with district court judges. Plus, we still had a long pipeline of EAs to draft. The pipeline included remaining projects under our first Notice of Funding Opportunity (NOFO), as well as dozens of projects under a second NOFO that we had issued for smaller supply chain investments. It was probably inevitable we’d fail to keep pace with deal flow.

Congress Intervenes

This is when Congress intervened, passing the Building Chips in America Act in October 2024. The bill exempted the majority of our projects from NEPA by carving projects with certain characteristics out of the definition of “major federal action.” (It didn’t exempt Micron New York — hence the government only recently completing its EIS.)

The bill was driven by Capitol Hill — the administration remained officially neutral — and was championed in particular by Senator Mark Kelly of Arizona and Senator Ted Cruz of Texas. In the Senate, it passed by unanimous consent; in the House, 178 Republicans and 79 Democrats voted for it, while 13 Republicans and 112 Democrats voted against. Though there was internal disagreement within the Administration as to whether the President should sign the bill, he ultimately did. Within Commerce, we had been tracking the legislation for a while and providing technical drafting assistance at the staff level, but I was shocked that it ultimately passed. It’s an example of how sustaining bipartisan support can shape outcomes in unexpected ways.

The shift to substance

The bill’s immediate impact was to substantially reduce (though not eliminate) the probability that NEPA would keep us from meeting our goals. It also changed internal administration dynamics. Before the bill passed, the overwhelming focus within the administration was on the NEPA process — for example, the White House in particular was understandably very focused on ensuring NEPA wouldn’t delay the program’s implementation. After the bill passed, the discussion shifted to substance: what were we doing to mitigate the environmental impact of the fabs we were funding?

We ended up negotiating the mitigations we’d planned to include in our EAs directly into our award contracts with companies. This included provisions around toxic chemical exposure limits for workers, reducing fluorinated greenhouse gas emissions, segregating and disposing of PFAS-containing waste streams, and recycling water. These were generally aligned with established industry practices and standards while accounting for individual operational and environmental considerations of each facility.

Would we have negotiated these provisions if NEPA had never been in place? It’s hard to say definitively, but I think we probably would have. Much of what we required was consistent with existing industry practice, so in many cases we were really cementing commitments, rather than breaking new ground. I’m personally proud that we included them, though the policy merits of including requirements like these in industrial policy awards can be debated — an issue tied to the “everything bagel” critique we’ll address later in Factory Settings.

What’s clear to me is that the cost of NEPA — that dual risk of delay and litigation — was vastly disproportionate to what it achieved. If the goal is to include environmental or other policy terms in awards, Congress has more straightforward tools. The CHIPS Act itself contained provisions related to workforce development, national security, and support for small businesses, none of which created the same combination of delay and litigation risk.

Closing reflections

I’ve spent my career in state and federal government in a range of roles, but if there’s one consistent thread, it’s been trying to show that government can work well and do some big things. At CHIPS, we set out to prove that the government could be nimble, commercial, and dynamic enough to negotiate with industry and onshore production of technology critical for national security. As a country, we’ll have to prove that again and again as our government takes a more proactive role in building resilience and competing with China.

What I struggled with most about NEPA was that, for nearly all semiconductor manufacturing projects in the United States, it only became a stumbling block when our government program got involved. For purely private investment, the array of federal, state, and local laws and regulations that govern construction and manufacturing are deemed sufficient safeguards for building a fab. That remains true if the company takes the investment tax credit from the IRS, which usually totaled more than double the grant funding from our program. But as soon as federal funding touched a project, a “major federal action” was created and NEPA kicked in — a massive undertaking with the looming threat of delay, litigation, and injunction.

To me, that’s a strange policy outcome: no NEPA for private investment; no NEPA for a large tax credit; yes NEPA for a smaller grant. The technical legal reason for this discrepancy is that the definition of “major federal action” excludes “non-discretionary” actions, and a tax credit is non-discretionary. And, to be sure, it certainly wouldn’t make sense to require environmental reviews for actions that the government is legally required to undertake regardless of environmental impact. But, in the context of industrial policy, the practical effect is perverse: NEPA creates a threat of failure precisely where the government is trying to be proactive, strategic, and engaged. For a program trying to prove that the government could operate with commercial speed and sophistication, that was a contradiction we struggled to resolve.

The CHIPS Program Office had a mammoth task: we had $39 billion to allocate and two years to do it. To get the job done, we’d need to negotiate with some of the most sophisticated companies in the world.

Startups often try to “move fast and break things.” But as a high-visibility, high-stakes federal program, we had to move fast but break nothing. To strike that balance, we needed to design a dynamic, outcomes-oriented process that gave us more discretion in decision-making and applicant engagement without sacrificing consistency or fairness. We quickly decided to update the typical grant procedure to emulate best-in-class private investment processes.

Our process was not seamless, and we certainly learned many things the hard way. But by the end of those two years, we had designed and stood up the new process, trained over 100 reviewers, and allocated $34 billion of the $39 billion to over 20 projects that advanced almost all the goals set in our Vision for Success (with $2 billion more allocated in preliminary term sheets)1.

Typical grantmaking wouldn’t have helped us meet our goals

The standard competitive grant review process is highly structured, formulaic, and “arm’s length,” limiting interaction with applicants. It’s also slow. Here’s how it typically goes:

1. The announcement: An agency publishes a notice of funding opportunity (NOFO) with clear guidelines on what is expected of the applicant and project, and what evaluation criteria will be used to review applications. Funding levels are often prescribed (e.g., X% of capital expenditures or Y% of total project cost).

2. The application: Applicants submit massive application packages with highly detailed project information.

3. The review: Review teams (which often include external experts) score applications. Evaluation typically follows a points-based rubric, with some consideration for other factors, such as available funding or strategic priorities.

4. The decision: The agency announces the recipients and funding amounts.

Throughout the process, interactions with applicants are usually minimal and highly structured to ensure equal treatment per the Administrative Procedures Act (APA). The scope of a project does not change after submission, and reviewers have little discretion in how projects are evaluated on the whole — it’s all about the rubric.

This process’s main advantage is that it is easily defensible: no one can claim the government played favorites or made “arbitrary and capricious” decisions, the APA legal standard that haunts all federal programs. But as discussed earlier in this series, we wanted to manage that oversight risk in a new way: we wanted to judge our success by our outcomes, not our process.

The stakes for failing to deliver were too high; we needed a results-driven process that would help us manage the complexities of our mandate.

Our program had to make investments in a highly volatile and rapidly evolving sector. We’d need a rigorous diligence process to ensure our investments would benefit the US taxpayer. We also needed to craft award agreements that aligned incentives, ultimately funding projects that would encourage a flywheel of investment and create a robust domestic semiconductor ecosystem.

Given our wide-ranging goals, we had a diverse cohort of applicants representing the entire semiconductor value chain. Some were large, publicly owned companies building $20B+ facilities to manufacture cutting-edge chips for AI, while others were small, privately owned foundries manufacturing legacy chips for defense purposes. Some were expanding and modernizing older facilities, while others were building new ones. Some were clearly commercially viable in a well-known and established market, while others were developing and scaling newer technologies. We needed the discretion to evaluate projects individually and as part of a portfolio. While we regularly asked ourselves whether we could have just gotten by with a traditional tax credit, that model wouldn’t have enabled our goals.

But borrowing from private equity would. A typical private equity diligence process takes a rigorous and holistic view of a company, reviewing the overall market dynamics, product or technology, company financials, risk and growth prospects to determine whether to make an investment, and if so, at what price.

Subscribe now

Taking a page from private equity

We ultimately landed on a new approach to industrial policy. Since our process was the first of its kind, we hired top Wall Street and industry talent who understood commercial diligence and deal negotiations as well as the intricacies of the semiconductor industry. We also spent significant time training the 100+ staff involved in evaluations. And for oversight purposes, we documented everything.

While still a four-part process, our revamp allowed us to move quickly, shape outcomes, and enforce accountability. Our process consisted of:

• An abridged statement of interest

• An optional pre–application

• Full application submission and review, followed by preliminary term sheet for meritorious applicants

• Due diligence and award

1. Statement of interest (SOI)

In this phase, applicants submitted a short project description to demonstrate their interest in the program. We ended up receiving over 690 in total. We used these submissions for planning, and not for evaluation.

Early, low-effort submissions helped fill the pipeline. The semiconductor supply chain is among the most complex in the world, and our investments had to cover it all. We were without a blueprint and had no idea how many companies would apply; the SOI phase let us identify potential gaps in our pipeline, which led to building a proactive “go-to-market” function that sought out companies and projects that had not expressed interest but that we wanted to recruit to apply. We travelled to Taiwan, Japan, Korea, and across Europe to boost awareness and interest based on pipeline gaps, ensuring we targeted foreign companies across the supply chain.

That global outreach helped us succeed. For example, we initially did not see any advanced packaging projects in our pipeline, despite being the next critical breakthrough needed to overcome the limits of Moore’s Law. The US had less than 3% of packaging on its shores (and barely any advanced packaging), so we reached out to the top foreign companies in that sector to encourage them to build in the US. We ultimately awarded funds for three advanced packaging clusters, and the sector is now growing rapidly in America.

Gauging interest early also helped us properly allocate our resources. SOIs gave us a sense of how many pre-applications and full applications we would receive each month, letting us plan resources (such as hiring and procurement of consultants) accordingly. Without our SOI data, we might not have hired as aggressively across our investment, strategy, and legal functions and would likely have had to slow down our review processes.

2. Pre-application (optional)

Next, applicants could opt to submit a high-level project overview and project financials. In this phase, the CHIPS team provided formal feedback on strengths and areas that needed more development, offering a recommendation on whether to submit a full application, revise the project before submitting a full application, or not submit an application at all. This phase allowed applicants to get feedback on their projects without the effort of submitting a full project package. It also set the stage and working relationship with our applicants, which was critical when we were navigating the toughest parts of full application negotiations down the line.

Pre-applications multiplied the return on CHIPS dollars and led to more ambitious projects. Our teams met regularly with applicants (often multiple times a week) to learn about their projects and offer feedback. In pre-application feedback letters, we offered guidance on project scoping and financing. We recommended changes to capital stacks to catalyze more private capital and reduce reliance on government funding. We also pushed applicants toward bolder aspirations, such as challenging them to build on faster timelines, add more capacity, or push to a more cutting-edge technology.

The pre-app phase also helped us set expectations. Our leading-edge applicants, for example, had initially requested over $70B in collective funding for their fabs, exceeding the $39 billion we had to allocate. In the pre-app phase we signaled that they should expect much less, and ultimately funded multiple clusters for ~$26B total2. Another applicant submitted a project for a new technology that seemed to be a minor improvement on an enterprise technology use case, which we indicated would be low priority for us; they returned with a proposal that would represent a total breakthrough in AI datacenter chip technology.

Pre-apps also let us manage our time and prioritize projects based on economic and national security impact and need for CHIPS funds. We flagged high priority projects for expedited review, provided feedback to some projects that were promising but needed work, and told low priority projects not to submit a full application. This helped us tailor our full application pipeline to high-priority projects that served our target outcomes. While we received 167 pre-applications, we told over 50 applicants not to apply. Most took our advice, and we ultimately cut our pipeline down to 92 full applications, a large portion of which were deemed meritorious. Early vetting enabled us to allocate our resources more efficiently and not waste them on low-value applications. It also saved the companies time, sparing them the effort of compiling a full project package if they didn’t have a chance at funding.

3. Full application (and PMT) phase

To formally apply for the program, applicants submitted a formal application with detailed project and financial information, including driver-based financial models.

Our internal deal teams conducted a holistic review to determine whether or not the application met our criteria, the most important of which were the project’s impact on economic and national security outcomes and its commercial and financial viability. The Investment Committee would then decide whether to move a given application to a Preliminary Memorandum of Terms (PMT, or short term sheet) or to deny the application. For projects moving to PMT, we conducted detailed modeling to determine the level of funding needed (e.g., within 5-15% of capital expenditures). In some cases, we suggested scope changes to push applicants further, such as proposing new facilities or products.

Our full application process enabled us to evaluate a diverse set of projects and build a strong, diversified portfolio of outcomes. Because we used a qualitative and holistic review process, our deal teams and Investment Committee were able to spend less time reviewing, debating, and tabulating scores, and more time discussing each project’s economic and national security impacts and underlying investment thesis. It also allowed our teams to account for the complexity and nuance of “economic and national security impact” and assess how well vastly different projects could meet our target outcomes.

We were also able to protect taxpayer dollars by reducing information asymmetries between government and the private sector. Our teams conducted assessments that were much deeper and more wide-ranging than the typical grant review process. We pursued detailed financial diligence to build independent views of company financials, analyzing market dynamics using internal expertise and external datasets. Based on the data, we crafted our own assumptions to test applicant claims. Our teams also engaged regularly across the industry, interviewing customers and investors on market and product dynamics, and engaging applicants directly to pressure test the project’s potential impact and viability. By building a process that let us engage freely with applicants and the market and leverage multiple sources in our review, we were able to make more informed decisions.

More information helped us rightsize funding for each applicant, stretching the impact of each CHIPS dollar. For applications we wanted to advance to PMT, our teams conducted bespoke sizing of awards. We wanted to make sure projects were economical for the least amount of CHIPS dollars and that we were catalyzing, not crowding out, private capital. We analyzed everything from internal project returns to the cost basis of alternative geographies, ultimately arriving at a proposed funding level and a view on how the applicant could shore up more private capital. For one chipmaker, funding was informed by the cost of doing business in Singapore; for another, it was more about the cost of manufacturing in Germany versus the US. Some projects only needed 5% in funding to clear their hurdle rates and remain globally competitive, whereas others, such as suppliers who were not able to access tax credits, needed upward of 35% of project funds. In some cases, we told applicants to identify additional sources of non-government capital before we would consider moving to a preliminary term sheet.

4. Due diligence and award

Once we agreed on a PMT with an applicant, we kicked off a commercial diligence process to surface and mitigate any risks, relying on external transaction advisors. Our attorneys and deal teams partnered with outside counsel to negotiate a final Other Transaction Authority (OTA) agreement with the applicant, which detailed specific milestones and operational outcomes tied to tranched funding.

Our award process enabled us to shape priorities for each applicant. As in previous phases, our teams pushed companies to think bigger: to build faster, onshore more capacity, and do more for workforce development. At this stage, we challenged some of the leading edge and advanced packaging players to commit to more fabs than they had originally planned, and to accelerate their production timelines. We also pushed several defense players to expand their products to commercial markets to ensure their commercial viability. We were able to tailor additional commitments based on how each applicant could help us achieve our target outcomes. For one legacy chipmaker, the highest priority was securing inventory and manufacturing capacity in the event of a potential shortage (as seen in the COVID pandemic); for another, it was ensuring chips made in Taiwan were onshored faster.

Our award documents aligned incentives with our applicants via milestone-based payments. To ensure our goals for the project were aligned with those of our applicants, we defined specific operational and technical milestones for each applicant, unlocking funding only when those outcomes were achieved. We calibrated those milestones based on the outcomes we needed from each applicant: for some, we wanted to see customer commitments to ensure success; for others, we wanted to see funding secured; for others, it was about hitting operational or technical milestones on an accelerated schedule. In this way, we were certain we would only spend money if our applicants were achieving the outcomes we had agreed on.

Taken together, our process looked a lot more like a private equity or investment process than a typical grant review process - and as a result, helped us achieve the outcomes we needed.

Reflections for future programs

There is much active debate about the extent to which federal investment programs like CHIPS can more effectively incentivize industry versus traditional tools like tax credits. Tax credits are simpler in many ways - they are neutral and can be easier to administer. But when the aspiration is bold and the industry needs to move in ways it has not previously, it can make sense to leverage direct investment programs.

In the end, any program should be judged based on its outcomes. The CHIPS incentives program — enabled by our process — has been uniquely able to shape outcomes and enforce accountability. We were able to challenge applicants to pursue more ambitious projects to meet our economic and national security goals. And embedded milestones made funding contingent on achievement, so we were more assured of the reward.

CHIPS also made a good case for how to judiciously grant discretion within government funding processes and allow for more dynamic private sector engagement. We benefited immensely from taking a more collaborative approach. By having discretion delegated to our review teams and Investment Committee, we were able to tailor project outcomes and more effectively negotiate to protect taxpayer dollars.

Our process was not perfect, and it was tough to roll out (more to come on that!). But it gave us the latitude needed to achieve our goals, serving as a good model for future federal investment programs.

The expansion of Washington’s role in the economy raises fundamental questions about the scope, goals, and execution of industrial policy. Are these interventions a costly distortion of markets, or a necessary correction when markets break down? Should the government ever “pick winners”? When, if ever, is industrial policy justified?

A heuristic for industrial policy

I have spent my career in the center of American capitalism and believe deeply in the power of markets, competition, and comparative advantage. But I’ve also seen the economy hobbled by uneven playing fields, (rationally) ignored externalities, and brittle supply chains. In high-stakes situations where market forces fail to incorporate national security imperatives, targeted industrial policy can work. But interventions should be rare, limited to when four tests are met:

• The industry is strategically critical.

• The industry is compromised in a way that creates vulnerability and systemic risk.

• The industry has calcified in a way that perpetuates resiliency failures, and cannot (or will not) self-correct.

• The problem is correctable by a bounded government intervention that creates a sustainable market structure.

Critical

Given the current tendency to label everything a national security emergency, we need to be precise about whether an industry is truly strategic and critical. Industrial policy is appropriate when the industry has direct bearing on (a) US military or intelligence operations, (b) the physical health and safety of Americans, or (c) key technologies of future global competition. Interventions should also focus on industries that produce foundational inputs to others — those whose failure would ripple across the economy. Rare earths, magnets, semiconductors, and even batteries fit that description, as they all underlie multiple end products.

Compromised

But importance alone isn’t enough to warrant intervention. Another condition for industrial policy is that the industry is structurally compromised in a way that makes the system inherently brittle or creates a potential chokepoint. That might mean concentration in a sensitive geography, dependence on a single company, or conditions that lock in the status quo. If an industry is critical but widely diversified across multiple companies and allied geographies, intervention is unnecessary.

Calcified

Markets can be highly efficient, yet dangerously and systemically vulnerable.

In classical economics, a market failure occurs when the free market fails to allocate resources efficiently, leading to suboptimal outcomes in pricing or output. Critical industries are exposed to a different failure mode — markets often fail to account for or price in national security risk. Over time, supply chain vulnerability gets built into the industry structure. Industrial policy should target these calcified structures in critical industries that markets alone will not fix.

Markets rarely price in geopolitical tail risk or the collective value of supply chain resilience; if a risk is low probability but high impact, investors tend to discount it entirely. Take Apple, which has total dependence on TSMC fabs in Taiwan. A single catastrophic event — an earthquake, or a Chinese invasion — could halt all iPhone production. Yet Apple’s stock price shows no discernible discount for that existential exposure. The market effectively assumes such an event would cripple the global economy anyway, so it prices it as zero.

There are also public good and coordination failures that prevent markets from self-correcting. The national security benefits of supply chain resilience accrue to society, not any single firm. No one company can justify diversifying its supply chain if doing so raises costs and erodes competitiveness, especially while rivals take no action. As a result, rational companies perpetuate a collective vulnerability.

Correctable

For industrial policy to make sense, there must also be a plausible corrective path that leads to a self-sustaining outcome.

Executing industrial policy is notoriously hard. It invites political capture, bureaucratic error, unintended consequences, and market distortions. Temporary subsidies often metastasize into permanence. Resources are finite; every dollar spent on industrial policy is a dollar not spent elsewhere, so we must spend it well.

Still, there are situations when targeted, time-bound intervention can break a calcified market structure and crowd in private capital. The goal is to use public leverage to reach a tipping point, after which scale, learning, and ecosystem effects create a new equilibrium that can stand on its own.

Leading-edge logic semiconductors: a bullseye for industrial policy

One critical objective for the CHIPS Act was to stand up at least two scaled, leading-edge (LE) logic ecosystems in the US — each with multiple fabs and the infrastructure, supplier network, and workforce to sustain them. This vision was laid out in our Vision for Success in February 2023; one year later, we set an ambitious goal to manufacture 20% of the world’s LE logic semiconductors in the US by 2030. At the time, the US produced 0%. Achieving it would literally redraw the map of this crucial industry.

Leading-edge logic satisfied all of our prerequisites for industrial policy.

First, the sector was critically important. Leading-edge logic chips — sub-5nm transistors fabricated with extreme precision — power the most advanced systems on earth. They are the compute engines for AI accelerators and data centers; the brains of our smartphones and telecom infrastructure; and critical components in satellites, cyber defense, and certain advanced weapons systems. They also function as general-purpose technologies, propelling advances in drug discovery, autonomous driving, and quantum computing. They are the foundation of our technological future and therefore a clear national-security priority.

To make matters worse, the industry was vulnerable and compromised by a dangerous geopolitical chokepoint. Before CHIPS, over 90% of the chips used in US production were sourced from a single company (TSMC) in a single place (Taiwan). And that geography is under fundamental geopolitical threat from the PRC. Korea’s Samsung supplied the remainder, but lagged behind at the most advanced nodes. This combination was key: the product was strategically critical and concentrated in one of the world’s most geopolitically volatile fault lines. If those same chips were available from a range of firms across multiple allied geographies, intervention might not have been necessary. But the geographic and corporate concentration compromised the whole industry and created an unacceptable single point of failure.

Markets had calcified around the failure and could not correct it. Was this simply the “natural” result of comparative advantage — i.e., did Taiwan have fundamental resource, labor, or cost advantages that could not be overcome? No, the structure wasn’t natural. It was path-dependent and could have evolved in other ways. The US dominated the leading edge from the 1990s to the early 2000s, but lost ground as compounding factors took hold: Taiwan carved out subsidies and established a national focus on the industry; US financial markets rewarded asset-lite business models and offshoring; escalating complexity and cost drove out most competitors; and, importantly, Intel made some well-documented missteps.

Observers often allege that chips are manufactured abroad because labor is cheaper in Taiwan. But labor arbitrage is not decisive today. One report suggests that in 1991, labor was over 40% of costs industry-wide; today, extreme automation and increased capital intensity has driven that closer to 13% at TSMC. These changes to the industry, along with scale and ecosystem effects (and TSMC’s strong management), reinforced each other until the market structure became calcified — locked into an “unnatural” state that private capital alone could not break.

All of TSMC’s major customers (Apple, NVIDIA, AMD, Qualcomm, Microsoft, etc.) recognize the danger of this dependency on TSMC. However, none were willing or able to change the structure on their own. No company will willingly raise its own costs or lose queue priority at TSMC for the sake of collective resilience. Financial markets likewise ignore the existential geopolitical tail risk. You’ll search in vain for the Taiwan risk discount in any of these valuations. In short, the market failed to account for resilience.

We believed we could sustainably correct this vulnerability. The underlying economics weren’t hopeless, just skewed. By closing the up-front construction-cost gap, we hoped to change the calculus on new US capacity, unlock private capital, and let market forces coalesce around a resilient and competitive equilibrium in the US. We had a few factors in mind:

• Cost structure: The largest cost differential between the US and Taiwan was in construction, not operations. Once built, ongoing operating costs should converge – labor costs are higher in the US, but energy costs can narrow the delta. If we close that initial gap to Taiwan and Korea, US fabs can be close to cost-competitive over time.

• Ecosystem effects: Taiwan’s advantage is a dense ecosystem of suppliers, packaging facilities, and skilled labor. As new US fabs get built (and with targeted supplier incentives and investment in workforce pipelines), clusters will form, costs will fall, and self-reinforcing scale will emerge.

• Strategic value: Even if a small cost gap remains, value can outweigh price. Proximity to customers, integration with US research institutions and their talent, and greater supply chain resiliency all carry tangible worth. The US needs to be value competitive, not cost competitive.

A helpful analogy: Texas Instruments (TI) manufactures most of its mature analog chips — which sell for <$1 — in the US. Despite fierce price competition with its many Asian and Chinese competitors, TI remains globally competitive through scale, automation, and ecosystem depth. If the US can compete in the oversupplied, low-price point analog world, it should be able to compete at the leading edge too where pricing power is higher and there are fewer competitors.

Projected demand growth also gave us confidence in our ability to shift the market. AI and advanced compute workloads are driving an explosive need for leading-edge capacity — projected to grow 69% by 2028. To meet this demand, many new fabs will be built somewhere — policy can influence where.

In a flat market, adding capacity is zero-sum and likely destabilizing and self-defeating. But in a growing market, shifting supply geography can be positive-sum. The existing supply/demand balance and projected growth matter — and for LE logic, they create positive tailwinds.

Industrial policy as disciplined statecraft

Narrowly applied industrial policy can strengthen resilience and reshape industry outcomes in ways the private sector alone will not.

LE logic satisfied the four C’s of industrial policy: the industry was critical, compromised, calcified, and correctable. But discipline is everything. This framework can help identify other industry segments to target, but we should err on the side of caution. Every market intervention comes with downsides, economic costs, and unintended consequences. Execution — deciding which tools to use, in what combination, for how long, and with what outcomes and accountability — is a delicate art.

When used sparingly and executed with rigor, industrial policy can be an instrument of national resilience. But industrial policy is a corrective, and not a replacement for markets. In rare cases when markets alone cannot secure the future we need, disciplined industrial policy is the best tool to restore balance.

So imagine my surprise when, in the first Inspector General report on our program, the IG scolded us for not having written a “comprehensive workforce plan.” In the title, no less.

The IG report both praised and chastised us. On the one hand, the IG acknowledged that we had surpassed our hiring goals for both the CHIPS Program Office (CPO) and the CHIPS Research and Development Office (CRDO), and outlined how the Office of Human Resources Management increased staffing to support our hiring surge. But the report put greater emphasis on our failure to document our workforce plan. The expectation was that we should waste precious capacity on documentation, rather than on the work of actually hiring folks.

Why was oversight dinging us for moving fast? Congress clearly wanted us to act swiftly to address a massive national security risk. If we had paused to write the workforce think piece the IG wanted, we never could have delivered the program. We saw this time and time again: many systems in government are designed to move slowly. The fear of waste, fraud, and abuse (and, let’s be honest, headline risk) is embedded in nearly every process. This manifests in requirements like detailed, a priori planning documents, and a stringent division of duties that breaks down tasks into tiny sub-units, without overarching ownership. For fear of favoritism — or even the perception of it — government hiring processes tend to prize risk management over both efficiency and outcomes. But CHIPS benefitted from some unusual structural advantages that other programs would benefit from replicating. And we were glad to take the oversight hit to get the job done.

Subscribe now

How we built for speed

Prior to my work on CHIPS, I had been through my share of federal hiring processes across multiple agencies. When I became the 7^th official CPO employee in January 2023, I was pumped to learn that we were playing with a strong hand that would let us hire quickly: (1) ample resources; (2) special hiring authorities; (3) the right organizational home; and (4) centralized tracking.

Resources

We had ample funding for hiring. While our administrative cap was the same 2% that many supplemental programs get, 2% of $39 billion goes quite far. Scale made all the difference. It meant we could hire as many staff as we actually needed without the usual tradeoffs between positions.

Funding made everything else possible. We could have done everything we list below and more, but without the money and headcount, it would be game over.

Authorities

We also had special hiring authorities, which let us move faster than most traditional hiring processes allow:

• CHIPS 25: Most significantly, the CHIPS Act designated an authority known as CHIPS 25, which allowed us to very quickly bring on 25 people at pay levels indexed to the Vice President’s salary — substantially higher than most federal pay bands.2 That authority was critical to attracting candidates who made two to four times more in the private sector. They were still taking a pay cut, but CHIPS 25 let us offer something slightly more palatable.

• Direct hire authority: We also secured Direct Hire Authority (DHA), and used it to our advantage. The Office of Personnel Management can grant a DHA when an agency faces a critical hiring need or a severe shortage of candidates. In practice, a DHA lets you skip the most rigid parts of the traditional competitive hiring process, letting HR and hiring managers quickly evaluate and select more suitable candidates.3 But the efficiency gains depend on how the authority is implemented: different agencies interpret the rules based on accumulated lore and risk aversion (sometimes due to a single bad audit finding). For example, some colleagues pointed out that DHA’s public posting requirements don’t include posting to USAJobs. We couldn’t take that shortcut because NIST required we go the normal USAJobs route.
  
  Rather than fight it, we adapted: we used “soft postings” on other sites (such as LinkedIn) to generate interest and identify strong candidates early. We then published the formal posting for a short window once we knew we had a solid pool. This strategy was compliant but let us move quickly.
  
  DHA was ultimately invaluable to CHIPS, but organizational baggage can shape what’s possible within the same legal authority.

• Excepted service: Our final prized tool was excepted service authority, which creates new categories of positions outside Title 5.4 New excepted service positions require the agency to justify why standard hiring rules don’t work for the proposed positions, OPM to approve the exception, and notice in the Federal Register. This differs from DHA, which operates as a procedural modification to the standard competitive process.5
  
  Once established, hiring for excepted service positions is about as efficient as it gets: the hiring manager submits a resume that meets the requirements of the job and required paperwork, the HR team processes it, and quickly you have a new employee. We thought carefully about how to allocate our 50 excepted service slots, using them where speed mattered most. We gave every leadership team member at least a handful of slots, but otherwise we prioritized the investments and strategy teams. We knew we’d need them onboarded well before applications from companies started coming in.

Working with OPM to get both direct hire and excepted service authorities took way longer than it should have. It required formal memos, escalation to the agency leadership, and multiple follow-ups. And we had to justify which position series and grades we wanted each authority for up front. Sometimes we’d be quite far along in designing a position, only to realize it didn’t fit into the DHA or excepted service series we’d been granted. We’d have to start over.

We were initially granted 75 DHA slots and 50 excepted service slots. By the end of summer 2024, we came close to running out of DHA, so we went back to OPM and eventually got another 200 slots. Without CHIPS 25, direct hire, and excepted service, we would have been dead in the water.

It helped that Secretary Raimondo herself was focused on personnel. She helped to close many key recruits and encouraged a risk-taking mindset. That expectation from the top informed our approach and sent a strong message to our agency partners: we needed to do whatever it took to bring people on board for this critical mission.

Choosing the right organizational home

We also benefited from strategic organizational placement. We were embedded into NIST, rather than established as a new bureau. This granted us two crucial advantages:

• NIST had the most competent operations team in the Commerce Department. Its senior ops leaders were experienced and savvy government operators. They understood the priority of our mission and helped accelerate our processes.

• NIST’s pay structure gave us more control in designing pay packages. Unlike many other agencies, which operate under the General Schedule (GS) system, NIST uses pay bands. That system provides just a bit more flexibility in setting pay at the outset. NIST also had more latitude in defining bonuses. They were still small by private-sector standards, but the system was more modern and progressive than GS. This made a meaningful difference in recruiting and rewarding our staff.

Centralized hiring tracking

Every hire involves multiple teams (the program team, HR, security, drug testing vendors, ethics, onboarding, etc.), each with a bespoke system, its own priorities, and no visibility into the processes outside of its own. Each team treats a hire like a transaction that they are processing to get it out of a queue.

In most agencies, hiring is fragmented across disconnected systems, causing delays and errors. Luckily, one of the brilliant NIST leaders designed a unique centralized hiring tracker for us. This was a very simple Excel spreadsheet with some embedded macros, but it tracked the entire hiring process — from position conception to posting to badging and everything in between. Our tracker let us see, almost in real time, where each hire was in the process.

Lots of organizations make hiring trackers, but in my experience they are owned by a single team (like HR). In that model, the owning team becomes the adjudicator of truth (which they do to their own advantage), damaging trust in the tool. Having a source-of-truth with shared ownership that was transparent to all program leadership (not just the HR staff) was invaluable. Shockingly, I hadn’t seen anything like it in my 15 years in government.

This visibility let us do a couple of things differently:

• Better planning: Instead of waiting for a hiring cert from HR and cobbling together availability, each team could anticipate when hiring might begin and block time on the hiring panel’s calendars for interviews.

• Better process sequencing: Visibility helped us see where we could parallelize steps, or change the sequencing across teams. We found that rather than passing a hire from HR to security to legal, we could save a bit of time by starting legal and security reviews simultaneously.

• Prioritization: Tighter coordination let us set shared priorities across all of the functional areas. We could tell a team to slow down processing on one hire if we needed to urgently redirect focus to a different position.

• Focus on outcomes: Most importantly, end-to-end tracking shifted every participating team’s focus to the end game: was there a person in a seat, with a computer and a badge, ready to work?

Despite its simplicity, centralized hiring tracking is decidedly not the norm. Future initiatives that need to scale fast would do well to copy this approach — identify the key tasks of every team involved, track them in a central place (even if this means moving it outside of your official systems), and have senior folks run frequent meetings to talk through the details, prioritize, and to ensure that everyone in the process is focused on outcomes. It takes some elbow grease and some grit, but this is absolutely replicable.

Subscribe now

The IG comes to call

Given our efforts, we felt confident when the IG announced their first engagement on whether we’d met our hiring goals. They came in skeptical, dug through our data (made easier by the tracker), and couldn’t ignore the fact that we had exceeded our goals. But they still wrote us up — not for failing to hire, but for not following OPM’s best practices. Specifically: “NIST did not develop a comprehensive workforce plan to meet its human capital needs.”

The report went on to scold us, noting that,

“Although CHIPS leadership recognizes the importance of workforce planning, evidenced by the requirement for applicants to submit workforce plans for CHIPS funding, CHIPS officials did not hold themselves to the same standard by completing their own workforce plan for CPO and CRDO.”

This was a fundamental misunderstanding of the task at hand: staffing the construction and operation of semiconductor facilities — enormous, permanent, physical infrastructure requiring highly technical employees, of which there are demonstrable shortages in the US — is an entirely different matter than a start-up government financial assistance program. And their findings did not acknowledge that we were able to move quickly because we had the flexibility to change who we hired and what the roles looked like. It was a perfect reflection of a system more focused on documentation than results.

We tried to make all of these points in a meeting with the IG staff. The OPM guidance might make sense once you’re up and running, or for planning giant physical infrastructure that requires thousands of very specialized employees. In those cases, you need to think carefully about the evolution of your workforce so you don’t get caught flat-footed. But slowing down to produce a comprehensive plan for staffing CPO instead of actually hiring people would have been insane. At best, we’d be guessing, and would have invariably gotten it wrong and been dinged by oversight anyway. And given the priority of the CHIPS program, slowing down likely would have earned the ire of our congressional champions on both sides of the aisle.

The IG critique became something of an inside joke. We were happy to take the hit if it meant we actually got something done. And if writing that report made the IG feel like they’d delivered something for their $5 million annual budget, well then… fine?

Unfortunately, getting dinged for not following ridiculous rules comes with the territory in government. Jen Pahlka’s recent piece on reforming the overseers hit home. And I read the Department of Education’s response to the GAO’s report on FAFSA implementation with some envy — they detail all of the missed nuance and impossible expectations, offering revised recommendations to improve the GAO’s report. We had considered drafting a similar response, but ultimately decided to forego the argument and maintain our focus on the task at hand.

Master your toolkit

In early summer 2023, the significance of what we accomplished became clear to me. We had just crossed the hundred-person mark that June and held our first all-staff meeting at the Patent and Trademark Office complex in Alexandria. Mike and I had just wrapped a listening tour across every team, and we shared their feedback with the group. The upshot was that people were really excited to be there. They were inspired by both the mission and the talent around them.

The energy in the room was unmistakable, and memorable moments from that day set the tone for the year to come: Mary Alex Smith characterizing the vibe as “like summer camp,” Hassan Khan nearly hurling a chair during a particularly intense round of semiconductor trivia, and Ayodele Okeowo winning our very first Star Award, a tradition that would carry on for the next two years. That afternoon, as we started collecting input on what would become our cultural norms, the results of the hiring surge were clear. We’d built the team that would go on to build CPO — an exceptional group who would do the impossible, and who, in so doing, became evidence of what government can accomplish when it optimizes for outcomes.

Our experience staffing the CPO team shows what’s possible. With the right authorities, ample funding, operational excellence, and the right mindset, the government can hire people quickly. But this raises the question of whether we want success to require endless exceptions, or whether to make efficient hiring the norm. We need to get better at designing processes that manage risk without suffocating outcomes. Policymakers: I’m not suggesting that we eliminate competitive hiring rules. But when designing high-stakes programs, you’ll need modified rules to get the job done.

While risk aversion is indeed a government-wide pathology, we don’t need to wait for generational HR reform to give programs more flexibility. I hope to see systemic change soon. In the meanwhile, roll up your sleeves and find a way to mobilize the tools available.