The AI arms race just got a price tag: $100,000 per overseas hire, and the leaders building the future didn't blink.
The Summary
- OpenAI, Anthropic, and Nvidia all filed more H-1B visa applications in Q2 2026 than a year earlier, even as other tech giants with larger workforces cut back.
- New government rules imposed a temporary $100,000 fee on applicants living overseas, adding massive friction to foreign hiring.
- The split reveals a talent war bifurcation: frontier AI companies are paying whatever it takes for specialized researchers while traditional tech firms retreat from the visa program entirely.
- Updated lottery rules now favor higher-salary positions, making the program even more concentrated among elite AI roles.
The Signal
When the Trump administration slapped a $100,000 fee on H-1B applicants living overseas, most tech companies saw a reason to slow down. OpenAI, Anthropic, and Nvidia saw the cost of doing business. All three increased their H-1B filings in Q2 2026 compared to the same period last year, applications that now include both new hires and renewals. The divergence from broader tech industry trends is stark: companies with much larger workforces are cutting back on the program at the exact moment AI leaders are doubling down.
The math matters here. A $100,000 fee per overseas applicant fundamentally changes who can afford to play this game. For a company hiring hundreds or thousands of workers, the fee becomes a budget crater. For frontier AI labs competing over a few dozen critical hires, it's barely a rounding error on their compute bills.
"Companies are cutting jobs while concentrating talent in smaller, more specialized teams."
The talent bottleneck in AI is real and getting tighter. We're talking about people who can train frontier models, optimize inference at scale, or push the boundaries of what these systems can do. That pool is global and tiny. The visa lottery changes that now favor higher-salary positions actually work in favor of AI companies, where six-figure salaries for machine learning researchers are table stakes.
The filing patterns reveal something else: AI companies are building different kinds of organizations than the tech giants that came before them. Meta calls them "pods." Business Insider describes them as "smaller, more specialized teams." The common thread is concentration. Instead of sprawling engineering organizations, you get tight units of exceptional people. Foreign-born workers with specialized skills become central to talent strategy, not replaceable cogs.
This creates a weird dynamic where visa restrictions that look tough on paper actually entrench the advantages of well-funded AI labs:
- $100,000 fee? No problem when you're raising billions.
- Lottery favors high salaries? Great, we pay top of market anyway.
- Process is uncertain and complex? We have immigration lawyers on retainer.
Meanwhile, the broader slowdown in H-1B filings across parts of the tech industry suggests that traditional tech companies, many still digesting layoffs from 2023-2024, are finding it easier to hire domestically or just hire fewer people overall. The policy meant to restrict immigration is instead concentrating foreign talent at the companies building the most powerful technology.
The Implication
If you're watching where AI is headed, follow the talent flows, not the press releases. OpenAI, Anthropic, and Nvidia paying six figures in fees per applicant tells you something about the scarcity and value of what they're hiring for. These aren't generic software engineers. They're people working on problems that maybe 1,000 people on Earth can solve right now.
For anyone building an AI career, the signal is clear: specialization is the trade. The companies paying $100,000 visa fees aren't looking for generalists. They want people who can do things almost nobody else can. If you're trying to break into frontier AI work, that's your constraint and your opportunity. Get good at something impossibly hard, or get comfortable being on the outside of where this technology gets built.