Research Scientist at AI Labs: H-1B and O-1 Sponsorship at OpenAI, Anthropic, and DeepMind 2026

AI labs are among the most active H-1B and O-1 sponsors in the US — here is how to position yourself, navigate the visa options, and land the offer.

You have a strong publication record, a PhD from a solid program, and genuine research chops in machine learning or deep learning. You've done your homework and know that US AI labs — OpenAI, Anthropic, Google DeepMind, Meta AI, Microsoft Research — are the places doing the most interesting science with the most compute. The only friction between you and those roles is the visa question.

The good news is that AI research is one of the best fields in which to be an international candidate seeking sponsorship. These companies are large, legally sophisticated, and built their research organizations in part on globally recruited talent. They have in-house immigration teams. They understand the process. What they need from you is to be a candidate they want badly enough to start that process — and to walk in understanding how the system works so you can manage the timing yourself.

Why AI labs are among the strongest H-1B sponsors

The biggest AI research organizations have consistently ranked among the top H-1B employers in the country, and the nature of research scientist roles makes sponsorship straightforward from a legal standpoint. A research scientist or research engineer at an AI lab is an almost textbook "specialty occupation" under USCIS standards — the position requires at minimum a bachelor's degree in a specific and relevant field, and in practice a master's or PhD is the norm. DOL's prevailing wage system has established benchmarks for these roles at Wage Level III and IV, which AI labs easily meet given their compensation structures.

For context on the broader landscape, the machine learning engineer H-1B sponsorship guide covers roles that frequently overlap with research positions, and the AI jobs that sponsor H-1B in 2026 gives you a broader sense of employer patterns across the industry.

H-1B path for AI researchers — the mechanics

Specialty occupation requirements

A research scientist role at an AI lab satisfies the specialty occupation standard under 8 CFR 214.2(h)(4) because the theoretical and practical application of highly specialized knowledge — machine learning, neural network architecture, reinforcement learning, probabilistic modeling — requires at minimum a relevant bachelor's degree. In practice, because your competition for these roles typically holds PhDs, the specialty occupation argument is usually clean. If your title is "Research Engineer" rather than "Research Scientist," the same logic applies as long as the duties involve research-level technical work, not just deployment or maintenance.

The H-1B Modernization Rule (effective January 17, 2025) codified deference to prior approvals on extensions and transfers, which benefits researchers who receive an RFE on a first petition and then extend or transfer — USCIS must defer to the prior approval absent material error or new information.

The lottery math and master's cap

Research scientists at AI labs almost universally hold master's degrees or PhDs. If your highest US degree is a master's or above from a US institution, you're placed in the master's cap pool — approximately 20,000 slots — before the remainder enter the general pool of approximately 65,000 slots. Historically, master's cap selectees have had notably better odds than general pool entrants. You can only claim the master's cap benefit for a degree from a US institution; a degree from a foreign university, regardless of prestige, places you in the general pool.

If you did your PhD in the US, claim the master's cap. If you did your PhD abroad but have a US master's, claim the master's cap based on the master's. If both degrees are foreign, you're in the general pool.

OPT and STEM OPT bridge

Most international researchers at AI labs begin on F-1 OPT and bridge to H-1B through the lottery. The typical sequence:

1. Degree completes in May or December
2. OPT starts within 60 days of program end (EAD card processing takes roughly 3-5 months, so apply early)
3. Register for the H-1B lottery in March of the relevant year (registration window is typically early March, ~2 weeks)
4. If selected, employer files full I-129 petition April 1 onward for an October 1 start date
5. If OPT expires before October 1 and petition is pending and selected, cap-gap extends status through September 30

Computer science, statistics, electrical engineering, applied mathematics, and most engineering sub-fields appear on the STEM OPT extension list. If your degree qualifies, your STEM OPT extension gives you 24 additional months — up to 36 months total — which provides runway for up to three lottery cycles. The 90-day unemployment clock (150 days during the STEM extension period) applies throughout, so a research role at a well-capitalized lab is critical. If you lose your position, the clock starts counting immediately.

LCA and prevailing wage at AI labs

Before filing your H-1B petition, your employer files a Labor Condition Application with DOL. The LCA must certify that you'll be paid at least the prevailing wage for your occupation in your work location. AI labs in San Francisco, New York, and Seattle operate in high-wage MSAs where prevailing wages are substantial — but so is actual compensation, so this is rarely a friction point at the major labs. The LCA is certified electronically by DOL within about 7 days under standard processing. Once certified, the employer files the I-129 with USCIS.

O-1A visa — the lottery-free path for researchers with strong records

The O-1A visa covers individuals with extraordinary ability in sciences, education, business, or athletics. For AI researchers, it's a genuine alternative to H-1B — not just a fallback. There is no annual cap, no lottery, no October 1 start-date constraint. USCIS must adjudicate within about 2-3 months standard or 15 business days with premium processing.

The O-1 visa complete guide for 2026 covers the full picture. For researchers specifically, the relevant evidentiary criteria under the O-1A standard are:

You need to satisfy at least three of eight criteria. For a research scientist with two or three years post-PhD at a major lab, the most accessible criteria are typically: judging others' work (peer review is easy to document), scholarly articles (your publication list), original scientific contributions (your citation count and influence), and employment at a distinguished organization (the lab itself). Many mid-career AI researchers can satisfy three criteria without extraordinary difficulty. The challenge is documentation — you need strong reference letters and well-organized evidence packages.

The O-1A is sponsored by your employer and requires a petitioner. The employer files an I-129 on your behalf. Importantly, you can file O-1A while on H-1B or while on OPT (and O-1 is not subject to the cap-gap limitations of the H-1B). Some researchers at major labs hold both an H-1B and an O-1 concurrently — this is legal.

For a detailed comparison of when to choose one path over the other, see H-1B vs O-1 for researchers and academics.

Comparing AI labs as visa sponsors

The labs differ in scale and legal infrastructure, which affects your experience as a sponsored employee.

None of the major private AI labs are cap-exempt. Universities (MIT CSAIL, Stanford AI Lab, CMU ML, Berkeley BAIR) are cap-exempt, which means an H-1B filed through a university bypasses the lottery entirely and can start at any point in the year. The tradeoff is that university postdoc or research scientist salaries are substantially lower than private lab compensation. Some researchers use the cap-exempt university path to secure H-1B status and then transfer to a private lab — this is legal and common. The transfer from a cap-exempt employer to a cap-subject employer does not require a new lottery entry because the worker has already been counted against the cap (been previously cap-counted is the legal standard).

See the cap-exempt H-1B employers guide for the complete rules on which institutions qualify.

The green card path from AI lab researcher roles

Researchers at AI labs who intend to pursue permanent residence typically have three viable green card pathways:

EB-2 via PERM

Most AI lab researchers who start the green card process go through PERM labor certification first. The employer files with DOL, demonstrating that no qualified US worker is available for the specific position. PERM processing takes roughly 8-18 months in normal conditions, longer if audited. After PERM approval, the employer files the I-140 immigrant petition. For researchers from countries other than India and China, EB-2 priority dates are typically current or near-current, meaning you could reach permanent residence in 3-5 years total from PERM filing. For India-born or China-born researchers, the EB-2 backlog extends to decades — alternative strategies (EB-1A, EB-2 NIW) become critical.

EB-1A Extraordinary Ability — self-petition

Researchers with strong records can self-petition for EB-1A, which bypasses both PERM and the annual cap. EB-1A has the same eight-criterion evidentiary framework as O-1A but a higher evidentiary bar — the standard is that your achievements place you among the small percentage at the very top of your field. For a research scientist who has made original contributions that have significantly advanced the field, EB-1A is realistic. A strong publication record (highly cited papers, top-venue publications, substantial peer review service) combined with external recognition (press coverage, named awards, invited talks) forms a competitive EB-1A package. Major labs' immigration teams are familiar with building EB-1A packages for research scientists.

EB-2 NIW — National Interest Waiver

For researchers whose work has clear national interest implications — AI safety, healthcare AI, climate AI, national security AI applications — EB-2 NIW is a viable self-petition alternative. NIW allows you to bypass PERM (but not the I-140 queue) by arguing that your work is in the national interest, that you are well-positioned to advance it, and that waiving the normal job requirement is in the national interest. The standard comes from the Matter of Dhanasar (2016) AAO decision. Researchers in AI safety and alignment specifically have had success with NIW given the explicit national interest framing around these sub-fields.

Step-by-step timeline for an international AI researcher targeting a US lab role

1. Final year of PhD (12-18 months before target start): Begin networking, cold-emailing researchers at target labs, presenting at workshops and conferences. Build your publication and citation footprint now — these matter for both O-1A and EB-1A later.
2. Apply for OPT EAD at least 90 days before your program end date. Processing takes 3-5 months; apply early or apply at the start of your final semester.
3. Submit H-1B lottery registration in March of your target year. Registration window is approximately 2 weeks; your employer registers on your behalf. No fee for the candidate. Master's cap vs. general cap election happens here.
4. Selection notification typically comes in late March. If selected, employer files full I-129 from April 1 onward.
5. If not selected, evaluate: (a) try next year's lottery on STEM OPT extension, (b) explore cap-exempt university position to secure H-1B status and transfer later, (c) build O-1A record and petition, or (d) explore alternatives (TN for Canadians and Mexicans, E-3 for Australians, H-1B1 for Chileans and Singaporeans).
6. After H-1B approval (October 1 start or premium processing approval date): begin PERM process if EB-2 green card is the goal. The earlier you file PERM, the earlier your priority date.
7. 3-5 years in (for non-India/China nationals): I-140 approval, adjustment of status filing, work authorization via EAD while waiting.

Common mistakes that hurt AI researcher visa outcomes

Assuming the offer automatically handles everything. Even large labs vary in how proactively they manage immigration. Ask explicitly during the offer negotiation what the immigration support looks like, whether they cover premium processing, and whether they have an in-house immigration attorney or use outside counsel. Clarify before you sign. See the post-offer due diligence guide for international candidates for the exact questions to ask.

Missing the STEM OPT 90-day unemployment window. If you're between roles — even briefly — the clock starts. Postdoctoral fellowships, consulting agreements, or research internships can keep you compliant. Document all authorized work during OPT meticulously.

Not starting OPT early enough. OPT can begin up to 60 days before program end and up to 60 days after. If you want a May start date and your program ends in May, you need to have applied months earlier. EAD card delays (3-5 months) can push your authorized start date well past your intended start.

Building an O-1A case without documentation. Many researchers qualify for O-1A on merit but can't demonstrate it because they never kept records of peer review invitations, award citations, or press coverage. Build the paper trail while it's happening — save peer review confirmation emails, citation counts at filing time, and every invitation to speak.

Ignoring the priority date math for green cards. If you're from India or China, the EB-2 backlog stretches to decades. Starting PERM on day one of employment, exploring EB-1A early, and understanding the EB-3 downgrade strategy (which can sometimes result in faster movement for India-born applicants) matters enormously for long-term planning. This is not a decision to defer — every year of delay is a year of priority date you're leaving on the table.

Treating lab affiliation as an automatic O-1A qualifier. Working at a prestigious lab is one of the eight O-1A criteria (critical employment), but one criterion alone is not sufficient. You need at least three. The lab affiliation helps, but it doesn't do the work on its own.

Frequently asked questions

Do top AI labs like OpenAI and Anthropic sponsor H-1B visas for research scientists?

Yes. OpenAI, Anthropic, Google DeepMind, Meta AI, and Microsoft Research are among the most consistent H-1B sponsors for research roles in the country. These companies routinely sponsor both new-cap and transfer petitions. Because AI research requires highly specialized skills, they have strong incentive to sponsor internationally talented candidates and invest in premium processing and legal support.

Should an AI researcher pursue H-1B or O-1 visa sponsorship?

It depends on your publications and external recognition. H-1B is the default path — it covers the vast majority of research scientists at US AI labs. O-1A is worth considering if you have strong publication records, citations, invited talks, peer-review service, or competitive prizes. The O-1A avoids the lottery entirely and has no annual cap, but the evidentiary bar is genuinely high. Many researchers start on H-1B and file O-1A concurrently or after building their record.

Does a PhD help with H-1B specialty occupation approval for a research scientist role?

Yes, substantially. Research scientist roles at AI labs require at least a master's degree and typically a PhD in a relevant field (computer science, statistics, mathematics, or a closely related discipline). A PhD strengthens the specialty-occupation argument and also qualifies you for the H-1B master's cap pool, which historically has had meaningfully better odds than the bachelor's cap pool in lottery years.

What is the OPT and STEM OPT timeline for AI researchers before H-1B?

After completing a STEM-qualifying degree (CS, statistics, electrical engineering, and related fields all qualify), you get 12 months of OPT. If your degree qualifies under the STEM OPT extension list, you can apply for an additional 24 months, giving you up to 36 months total. During STEM OPT you must have a formal training plan (Form I-983) with your employer and must not go more than 90 days unemployed in total (150 days in the STEM extension period). Plan your H-1B lottery application timing carefully within this window.

How does the H-1B cap-gap provision protect AI researchers?

If your OPT expires on or after April 1 and your H-1B petition was selected in the lottery and is pending, cap-gap automatically extends your OPT status and work authorization through September 30 (or through the H-1B start date if earlier). Under the H-1B Modernization Rule effective January 17, 2025, cap-gap protection now extends through April 1 of the relevant fiscal year rather than the prior October 1 endpoint.

Navigating the H-1B lottery, O-1A, and green card path at AI labs takes careful timing — and the details matter. Reach out to F1Jobs to talk through your specific situation. We work with international researchers in AI and machine learning every month.