In August 2022, the United States passed the CHIPS and Science Act, committing $52 billion of direct federal subsidies, a 25% investment tax credit, and approximately $200 billion in research authorizations to the project of rebuilding domestic semiconductor capacity. The Inflation Reduction Act followed three months later, with another $380 billion oriented around domestic battery manufacturing, clean steel, hydrogen, and grid infrastructure. By the most conservative tally, the federal government has committed roughly $640 billion to the project of reshoring strategically significant manufacturing.

In the same window, the H1B cap has remained fixed at 65,000 specialty occupation visas annually, with an additional 20,000 for graduates holding U.S. master’s degrees or higher, numbers set by statute in 1990 and not adjusted upward since. Optional Practical Training (OPT) has been litigated continuously since 2018. The EB2 employment based green card category runs to roughly six years for Indian nationals applying today. The State Department’s consular discretion authority has been exercised at approximately 30% higher rates than 2019 levels for applicants from China.

These are two industrial policies. They do not agree. More fundamentally, they share a defect: both assume the federal government has the institutional capacity to centrally direct a labor market it does not understand. The first policy says: pay companies to build factories here. The second says: keep out the lawful, vetted, high skill workers those factories would hire. The contradiction is not an accident. It is the structural consequence of administering the economy as if it were a planning problem.

A clarification is owed up front, because the subject invites misreading. Nothing here concerns border security or illegal entry, which are separate questions with their own legitimate answers. The argument is narrower and, we think, harder to dispute: that the lawful, merit based, employer sponsored, security screened immigration of the world’s best engineers is a national asset, and that the United States is currently treating one of its greatest competitive advantages as a liability.

What CHIPS and IRA Promised

The reshoring program is, by 2026 standards, roughly on schedule on capital deployment. Of the roughly $640 billion announced through the end of 2025, approximately $270 billion is in some active phase of construction or commissioning. TSMC Arizona Phase 1 has produced its first wafers. Intel Ohio is approximately on track for Phase 1 fab activation. The battery gigafactory build out across the Southeast is proceeding faster than most pessimistic 2022 forecasts allowed.

The bottlenecks are not where most observers expected. Permitting has been faster than projected. Capital availability has been adequate. Construction labor has been tight but manageable. The bottleneck is operations: who runs the plants once they are built.

The CHIPS Act, in its workforce provisions, authorized roughly $200 million for workforce development, approximately 0.4% of the bill’s headline funding. The Department of Commerce conceded in its 2024 implementation review that the workforce shortfall would constrain operational ramp by 18 to 36 months at most subsidized facilities. The remediations on offer are apprenticeship program expansion and community college partnerships. These are good. They are also a decade slower than the timing the bill itself contemplates, and they reflect the same federal program thinking that produced the labor shortage in the first place.

The unspoken assumption inside the workforce development discussion has been that the labor will come from somewhere. The discussion has not addressed where. Historically, and in the United States decisively, a large share of it has come from skilled immigrants who arrived legally and stayed.

What H1B and OPT Withhold

The H1B cap was reached, in 2024, in less than a single day. USCIS received 470,342 registrations for 85,000 slots, an applicant to slot ratio of approximately 5.5 to 1. The visa lottery, instituted in 2007 to manage oversubscription, randomly distributes work authorization almost regardless of employer, role, or strategic relevance to American industrial policy.

A semiconductor PhD employed at TSMC Arizona is entered into the same lottery as a contracting house mid level developer who will be staffed to a client in suburban New Jersey. The State Department has no discretion to weight the allocation toward strategic sectors. Commerce has no instrument for indicating that a specific applicant works in a CHIPS subsidized facility. The two policy regimes operate as if they belong to different governments.

The two policy regimes operate as if they belong to different governments. In a sense, they do.

OPT compounds the contradiction. Foreign students at U.S. universities can, under current rules, work in the United States for up to three years after graduation in a STEM field. This program is the single largest source of skilled immigrant talent entering American industry. It graduates approximately 240,000 STEM OPT participants annually. Of those, approximately 60% are estimated to leave the country at OPT expiration because they cannot secure H1B sponsorship in time. They take their U.S. subsidized graduate degrees and they staff foreign competitors.

A rough estimate: between 2010 and 2024, the United States educated approximately 1.4 million STEM graduate students at its universities at a total federal subsidy cost of roughly $40 billion. Of those, an estimated 720,000 left at OPT expiration. The single largest direct beneficiary of American graduate STEM education, by simple count of retained graduates, is the People’s Republic of China. This is not a hypothetical industrial policy failure. It is the documented one, and it is, at root, a national security failure: we educate our chief strategic competitor’s technical workforce and then escort it home.

A Brief History of Government Getting in the Way

It is worth establishing, against the conventional framing, that the United States has, for most of its history, been a country whose industrial leadership rested on the lawful recruitment of exceptional people. The framing in which immigration is one debate and industrial competitiveness is another is roughly forty years old. It is also wrong.

The first wave was the postwar physicist diaspora. Roughly 1,500 European trained physicists, mathematicians, and chemists arrived in the United States between 1933 and 1948. The visa regime that admitted them was, by current standards, extraordinarily welcoming toward vetted technical talent. The State Department, working under Roosevelt and Truman directives, treated technical talent recruitment as a strategic priority and allocated visa capacity accordingly. The contribution to the American industrial and defense base does not require restatement.

The second wave followed the 1965 Hart Celler reform that abolished the country of origin quotas. The cohort of Indian and Chinese engineering graduates who arrived in the United States between 1968 and 1985 is, in the aggregate, among the most productive immigrant generations in American industrial history. Half of Silicon Valley’s senior leadership in 2025 is, in some sense, a downstream consequence of a 1965 policy choice.

The third wave, the H1B program enacted in 1990, was structured around the assumption that the United States would continue to be a lawful importer of technical talent. The program has done that work, intermittently, but the cap was set at a level that bore no relationship to the labor demand it was meant to meet and has not adjusted in a generation. The semiconductor industry’s leadership team, across the major American firms, is approximately 40% first generation immigrant. This is not a fact about Silicon Valley culture. It is a fact about whether the country was willing, at the relevant moment, to admit talent through the front door.

The current regime, in which a country pursuing a trillion dollar industrial policy simultaneously restricts the lawful inflow of the workers required to run the resulting capacity, is anomalous in American history and incoherent on its own terms.

The PhD Retention Question

The most consequential and least discussed dimension of the contradiction is what happens to Chinese national PhD students at U.S. universities.

Approximately 35% of the PhDs awarded in American semiconductor engineering, materials science, and applied physics over the past decade went to Chinese national students. A roughly equivalent share of AI and machine learning PhDs. The 2020 Presidential Proclamation 10043 began the formal effort to restrict Chinese national graduate students in specific sensitive fields, and subsequent administrations continued the restrictions in modified form. The net effect is that approximately 18,000 Chinese national PhD students per year, graduating from American programs at the top of their fields, with skills directly relevant to the technologies the United States has subsidized at industrial scale, face a degraded path to remaining in the country.

The strategic logic of the restriction deserves to be taken seriously rather than dismissed. There are documented cases of technology transfer to mainland China through return trip flows of Chinese trained American PhDs, and the concern that motivated the now disbanded China Initiative was real. But the honest accounting is that the current approach achieves the worst of both outcomes. A Chinese national PhD in advanced packaging at Purdue, who would have stayed and worked at Intel Ohio for the next twenty five years under a sane and security screened pathway, is now disproportionately likely to return to Shanghai and work at SMIC. The technology transfer happens anyway, because the knowledge is in the person, who carries it home at OPT expiration. We have built a regime that pays to train our competitor’s workforce and then guarantees its return.

This is, in our judgment, among the largest unforced errors in American industrial policy of the past decade. The correct response is not less screening but more intelligent screening: rigorous, individualized security vetting at the point of degree completion, followed by a durable status for those who clear it. That is a policy that serves national security and industrial competitiveness at the same time, rather than sacrificing both to bureaucratic momentum.

A reformed regime would do close to the opposite of the current one. The proposals that emerged from the 2024 Congressional Hard Tech and Talent Working Group included a durable status for security screened STEM PhDs graduating from American research universities, with the review applied at degree completion rather than as a permanent barrier to retention. This is precisely the kind of reform that closes the contradiction. It has been proposed in various forms since the late 2000s. It has never been enacted, because the coalition that would enact it, security minded supporters of lawful high skill immigration, has not yet organized itself as a constituency.

What the Comparators Do

The United States is not the only country attempting to navigate industrial competitiveness in an era of contested talent policy. It is, however, the only one we are aware of with a coordinated industrial policy that is materially undercut by its own immigration rules.

Canada operates the Express Entry system, which scores skilled migrants on a transparent ranking and admits roughly 110,000 federal economic class permanent residents annually. The system is fast, with a six month median, employer blind in a way that makes it less vulnerable to corporate capture, and explicitly oriented around long term economic contribution. Roughly 28% of Canadian software engineering hires in the past five years entered through this pathway. Toronto, Vancouver, and Waterloo have absorbed substantial talent flows that the American system has, in effect, exported.

Australia operates the Global Talent Independent program, capped at 5,000 grants annually, trivially small in absolute terms but explicitly targeted at sectors the country has identified as strategic. Adjudicated cases generally close in under three months. The targeting list is set by the Department of Industry, Science and Resources, meaning the visa program is, in fact and in law, an instrument of industrial policy.

The United Kingdom operates the Global Talent visa, the High Potential Individual visa, and the Scale Up visa. The framework explicitly conceptualizes lawful skilled immigration as part of the country’s industrial strategy and is administered by the Home Office in consultation with the Department for Business and Trade.

Each of these regimes is imperfect by the standard of a genuinely open market for talent. Each grants the executive substantial discretion. None of them is the right model in the abstract. All of them are more coherent than the current American regime, because each treats lawful skilled immigration as a strategic asset rather than as a separately contested political question, and each accepts that a country’s industrial ambitions have to be matched by its willingness to admit the people who will realize them.

Three Specific Reforms

We are not in the business of policy advocacy, but the implications for the talent acquisition function inside American industry are direct enough that boards and operating teams should understand them. Three reforms would close the contradiction without requiring any compromise of border enforcement or the rule of law.

  1. A CHIPS Act sectoral visa allocation. Reserve a defined fraction of the H1B cap, perhaps 20,000 of the existing 85,000, for applicants whose petitioning employer is a recipient of CHIPS or IRA subsidies. USCIS already tracks employer relationships in the petition data. It is the minimum coherence the federal government’s two industrial policies should be expected to achieve, and it requires a rulemaking, not a statute.
  2. A durable, security screened status for STEM PhDs. Implement the proposal that has been in front of Congress, in some form, since 2009, paired with rigorous individualized vetting. The cost benefit analysis is overwhelming and has been overwhelmed by inertia.
  3. Eliminate the per country green card cap. The current statute caps employment based green cards at 7% per country of origin, with the predictable result that the line for Indian nationals, who form the largest single share of STEM PhD recipients, runs to multiple decades. It is the most indefensible artifact in the current system, and it does no work that any serious security analyst would defend.

The deeper observation, beyond any specific reform, is that the contradiction between subsidy and restriction is not stable. A program of trillion dollar industrial subsidies that cannot lawfully acquire its own workforce is going to fail, fairly visibly, at some point in the next ten years. When it does, the political alignment will shift, and one of two things will happen. The country will either modernize lawful skilled immigration in the manner suggested above, or it will substantially curtail its industrial ambitions. Both outcomes are coherent. The current intermediate position, building the plants and keeping out the engineers, is not.

What We Tell Clients

In the meantime, the operational guidance to American industrial employers is unsentimental. Lawful visa sponsorship infrastructure is now a material competitive asset and should be developed as such. The companies positioning themselves best are those that have built in house immigration counsel, structured operating team rotations through Canadian and British subsidiaries that provide lawful pipelines back into the United States, and built explicit succession plans around credentialed talent that has cleared security review. This is unusual practice in 2026. It will be standard practice by 2030.

We will close with one observation. The country that won the industrial leadership of the 20th century was the country that, in 1933 and 1948 and 1965 and 1990, made the deliberate choice to admit exceptional people through the front door, lawfully and on the merits. The country now subsidizing the project of recapturing that leadership has, at the same time, made it harder for exactly those people to come and to stay. The two projects cannot coexist. The resolution does not require open borders or the abandonment of any principle a serious person holds. It requires only that the country decide its legal immigration rules should serve its national interest, and then write them that way.

The market would solve most of this. The country has chosen, so far, not to ask it.