The letter
An Open Letter on USCIS PM-602-0199 and the Future of American Talent
The point
The United States leads the world in AI. This administration has the opportunity to lock that lead in. The lead is also narrower than it looks. For nearly a century the United States has compounded one structural advantage no competitor has matched: it has been the destination of choice for the world's top scientific and engineering minds. That single input, more than any other, is why American companies set the AI frontier today, why American universities anchor global research, and why American GDP per capita is what it is. Competitors understand this and are investing heavily to close the gap, on the talent input most of all.
Closing loopholes in adjustment of status is directionally correct. The system is abused and tightening it is the right call. But as written, PM-602-0199 will hit the wrong population hardest: the O-1, EB-1A, EB-1B, and EB-2 NIW scientists, engineers, and founders who built more than half of America's billion-dollar companies and who keep the United States ahead of China in AI.
This letter is not an objection to the reform. It is a request for two narrow, grandfathered carve-outs, and an invitation to design a 2026 talent program together.
On American workers and national security
We want to address two concerns upfront, because they should frame everything that follows.
American workers come first. Nothing in this letter substitutes for investing in the American STEM workforce. We are members of that workforce and we hire from it. U.S. graduates, U.S. apprenticeship pipelines, U.S. community colleges, and U.S. national labs are the foundation. Skilled immigration is additive: each immigrant-founded unicorn creates an average of 859 jobs, the overwhelming majority filled by Americans (NFAP, 2022; Forbes). The pipeline we are asking the administration to preserve has created millions of American jobs and trillions of dollars of American value, and it must be preserved alongside, not instead of, domestic talent development.
Counterintelligence vetting should be thorough. Nationals of adversary states working on dual-use technology require rigorous screening. The Visas Mantis program and the Technology Alert List exist for good reason, and we support them. We are not asking for lower standards. We are asking that security review run in parallel with status, not in place of it: keep the scientist working at a U.S. lab under U.S. oversight while the check completes, instead of sending them abroad where U.S. visibility ends and a foreign competitor can hire them on the spot.
What is at stake
- 55% of U.S. billion-dollar startups have at least one immigrant founder. 319 of 582 unicorns (NFAP, 2022). Combined value over $1.2 trillion. Average of 859 jobs per company.
- Roughly one in four (143 of 582) had a founder whose pathway ran through H-1B, O-1, or an employment-based green card, the exact statuses PM-602-0199 now restricts.
- A disproportionate share of foundational papers in modern AI, biotech, and physics come from researchers who entered the U.S. on O-1, J-1, or H-1B and naturalized through EB-1A or EB-1B.
Why AI talent is different from past technology races
Nuclear, semiconductors, and biotech each compounded fast, but every leader could eventually be caught. Many in our community believe AI is the first technology where that may stop being true. The leading frontier labs are explicitly building toward systems that use their own outputs to design the next generation of training, hardware, and algorithms. Whether or not full recursive self-improvement is years or decades away, the trajectory is steep enough that the gap between first and second compounds rather than closes. One exponent cannot catch another exponent.
The American lead in AI today is real. It is also small enough that the difference between first and second can plausibly be decided by which country a few hundred specific scientists choose to live in over the next two to three years. That is the population PM-602-0199, as written, is most likely to push abroad.
Where PM-602-0199 breaks
Under prior practice, an O-1 scientist or founder filed Form I-140 (EB-1A or EB-1B), obtained approval in roughly 15 days under premium processing, and filed I-485 to adjust status inside the U.S. while continuing to work. PM-602-0199 directs USCIS officers to deny I-485 absent "extraordinary circumstances" and routes the applicant to consular processing through the Department of State.
For most of the world this is inconvenient. For STEM applicants from a specific set of countries, it is a multi-year stop:
- No U.S. visa services in their home country. Embassy Moscow no longer issues non-diplomatic visas; Russian immigrant visa cases are routed to Embassy Warsaw with limited capacity. Iran, Syria, Venezuela, and Cuba have no U.S. visa services. Ukrainians and Belarusians are processed through third-country posts. "Return home to apply" is, in many cases, procedurally impossible.
- Visas Mantis administrative processing under INA §221(g). Applicants in AI, biotech, aerospace, nuclear, and advanced materials are flagged under the Technology Alert List and held for Security Advisory Opinions under INA §212(a)(3)(A)(i)(II). Nationals of Russia, China, India, Iran, Pakistan, and Israel get this treatment by default. Current wait: 8 to 18 months (State Department).
- No bridge work authorization. When the O-1 expires while the applicant is abroad, work in the U.S. stops. The U.S. employer (often Google, OpenAI, Anthropic, a national lab, or a U.S. university) loses a key contributor for the duration. At that point a recruiting call from Beijing, Abu Dhabi, or Paris with a job that starts Monday becomes the rational choice.
Concrete example: a Russian AI researcher with an offer from Google DeepMind in Mountain View
- Cannot apply at home. Embassy Moscow is closed for visa services. Embassy Warsaw is the State Department's designated processing post for Russian residents, but Poland does not issue visas to Russian nationals for the purpose of attending a U.S. consular appointment, which makes the designated route effectively unreachable for most applicants.
- UK Global Talent visa as the practical bridge. The applicant applies for the UK Global Talent visa (AI track, endorsed by UKRI or Tech Nation), moves to London, starts at DeepMind UK. From day one the UK gets the labor, the taxes, and the IP.
- Files I-140 from London. Approved in ~15 days under premium processing. Schedules an immigrant visa interview at a third-country consulate.
- 221(g) refusal under Visas Mantis. Because AI is on the Technology Alert List and the passport is Russian, the consular officer issues a 221(g) refusal pending a Security Advisory Opinion. 8 to 18 months, no U.S. work authorization during the wait.
- By month 24-36 the move is no longer realistic. Family in UK schools, UK equity, UK social ties. DeepMind London grows. Mountain View does not.
What we are asking for
Immediate: two narrow, grandfathered carve-outs
These are limited to cases already in the pipeline as of the effective date of PM-602-0199. They preserve reliance interests without weakening the prospective reform:
- Grandfather I-485 applications filed on or before May 21, 2026. Applicants who were lawfully in process before the new memo should be adjudicated under the prior framework. This is a standard reliance-interest carve-out and is consistent with how USCIS has historically handled mid-stream policy changes.
- Treat an approved I-140 in EB-1A, EB-1B, or EB-2 NIW that was filed on or before May 21, 2026 as a presumptive extraordinary circumstance for I-485 adjudication, where the applicant is in lawfully maintained nonimmigrant status. USCIS has already adjudicated extraordinary ability, outstanding research, or national interest in these specific cases. Forcing them back through consular processing now adds no security value and creates the operational risk described above.
We also urge State and DHS to coordinate on a Mantis fast-track for already-approved EB-1 and EB-2 NIW petitioners: same vetting standard, parallel processing instead of sequential.
Strategic: a 2026 American AI Talent Initiative
This administration is uniquely positioned to do something historic. Three times in the last century, deliberate U.S. policy concentrated the world's best technical talent on American soil and reshaped the next fifty years: Operation Paperclip in 1945, the Immigration Act of 1990, and the Soviet Scientists Immigration Act of 1992 (Appendix A). Each was the work of a specific administration that chose to act when the strategic window was open. AI is the equivalent moment in our generation, and the window is open right now.
The same statutory authorities that produced those wins are available today, with AI in the role nuclear played in earlier decades. The U.S. tech community is offering to help build and operate this with the administration:
- A dedicated annual allocation of EB-1A, EB-1B, and EB-2 NIW capacity for applicants in AI, biotech, advanced manufacturing, energy, and quantum: the modern Technology Alert List inverted from a watchlist into a target list, with vetting standards held constant or raised.
- A Mantis fast-track for security review of EB-1 and EB-2 NIW approvees, run jointly by DHS, State, and ODNI, with no reduction in scrutiny.
- Direct outreach to scientists currently stranded in third countries by the closure of U.S. posts in their home countries, and to those weighing offers from China, the Gulf, and Europe right now.
The undersigned offer this community's time, networks, and resources to identify candidates, screen them in coordination with U.S. authorities, and place them at U.S. labs, universities, and companies. Many of us already do this work informally for the people we hire and fund. We can do it at scale, in partnership.
Closing
The reform is correct. The loopholes should close. We are asking only that the front door stays open at the same time, for the population the United States has always benefited most from welcoming, and at a moment when the technology in play may not allow a second chance.
The opportunity is bigger than at any point in living memory. The cost is lower. The talent is already at the door.
We are ready to work on this together.