Anthropic just scored major win in escalating Al talent war
In Silicon Valley, the same handful of people keep showing up on opposite sides of the same fight. They build a company. They leave. They build another one. They poach each other's deputies. Over a decade, the alumni network slowly turns into a battle map, and you can read the state of an industry by tracking which way the marquee names are walking.
That dynamic has shaped every major tech race I have covered, from the early cloud wars to the autonomous driving land grab at Tesla (TSLA), Waymo, and Cruise. The pattern is always the same. Money matters. Mission matters more. And when one company starts pulling rival founders, the prediction markets usually catch on before the press releases do.
Right now, the AI industry is running the loudest version of that cycle in its short history. OpenAI's growth has been spectacular, but its alumni list has quietly become a recruiting pipeline for its biggest rival. Anthropic was founded by OpenAI exiles in the first place, and over the past two years it has kept pulling senior names away from Sam Altman's lab.
On May 19, that pipeline added arguably the biggest name yet. Andrej Karpathy, one of the original 11 founding members of OpenAI and the former head of Tesla's AI division, announced on X (formerly Twitter) that he is joining Anthropic. He will work under Head of Pretraining Nick Joseph and lead a new team focused on using Claude to accelerate pre-training research itself.
That last detail matters more than the announcement.
What Karpathy's move means for the AI race
Pre-training is the expensive part. It is the phase where massive models absorb language, reasoning patterns, and the rough shape of human knowledge before any fine-tuning happens. Small decisions made in that phase compound across every product built on top of the model.
Karpathy is one of the few researchers who can speak fluently in both the theory and the engineering reality of large training runs.
More AI:
- Micron sits at the center of a red-hot chip rally
- IBM CEO sends blunt message on AI and quantum computing
- Anthropic CEO makes shocking admission about AI
"I think the next few years at the frontier of LLMs will be especially formative," Karpathy wrote in his announcement on X. "I am very excited to join the team here and get back to R&D."
The corporate confirmation came from inside Anthropic. "He'll be building a team focused on using Claude to accelerate pretraining research itself," Joseph wrote in a post welcoming Karpathy, according to VentureBeat. "I can't think of anyone better suited to do it."
That second sentence is the actual strategy. Anthropic is not just adding a marquee hire. It is betting that AI-assisted research, where a model helps train its successors, is how it stays ahead of OpenAI and Google's Gemini team. The hire is "a clear sign from Anthropic that it believes AI-assisted research, rather than pure compute, is how it stays competitive with OpenAI and Google," reported TechCrunch.
Compute still matters. But if every frontier lab has access to the same chips and roughly the same capital, the differentiator becomes how cleverly you use what you have. That is what Karpathy was hired to build.
Related: Anthropic CEO makes shocking admission about AI
The OpenAI to Anthropic talent pipeline keeps flowing
What struck me when I mapped the flow over the past two years is how one-directional it has become. Karpathy is the third high-profile OpenAI figure to land at Anthropic in under two years, according to Benzinga.
Jan Leike, OpenAI's former head of alignment, defected in May 2024. Co-founder John Schulman followed that August. No comparable Anthropic researcher has moved the other way in that window.
That movement is happening against a backdrop of compensation packages that would have looked deranged five years ago. Anthropic's median total compensation for software engineers sits at roughly $600,000, with senior engineers earning a $316,000 base plus $247,000 in stock, according to compensation tracker Pin. OpenAI L5 engineers pull around $1.15 million in total comp, with senior individual contributors clearing $1.28 million at the top of the band, the same report found.
If money were the only variable, the senior talent would be moving toward the highest bidder. Instead, it is keeping going to Anthropic.
A SignalFire study highlighted by Fortune found Anthropic retains 80% of two-year hires while paying meaningfully less than OpenAI at the median. Culture, mission and team quality keep beating raw cash at the top of this market.
The OpenAI to Anthropic pipeline since 2024:
- May 2024, Jan Leike, OpenAI's former head of alignment, joined Anthropic after publicly criticizing OpenAI's safety culture, according to Benzinga.
- August 2024, John Schulman, an OpenAI co-founder, left for Anthropic, according to CNBC.
- May 2026, Andrej Karpathy, a founding OpenAI member and former Tesla AI director, joined Anthropic's pre-training team.
Why this hire matters for AI stock investors
Most of you cannot buy Anthropic shares directly. The company is still private and reportedly in talks for a roughly $30 billion funding round at a $900 billion valuation, according to Benzinga. That would surpass OpenAI's $852 billion mark from its March raise, per the same report.
But the secondary effects of this talent war are already showing up in stocks you can buy. Nvidia (NVDA) reports earnings on Wednesday with high expectations for AI demand, according to Yahoo Finance. Microsoft (MSFT), which holds a 27% stake in OpenAI under the April restructuring, has watched Anthropic's commercial momentum closely. Alphabet (GOOGL) has been positioning Gemini against both labs.
The Polymarket prediction markets are flashing a clear signal. Traders assign Anthropic a 70% probability of having the best AI model at the end of June, based on the Chatbot Arena leaderboard, with OpenAI given just a 5% probability, according to Benzinga. The same prediction markets give Anthropic a 67.5% chance of going public before OpenAI.
Those numbers were already in the cards before Karpathy's announcement. The risk for any investor holding indirect AI exposure through Microsoft is that Anthropic's research lead widens enough to start pulling enterprise contracts. KPMG just announced it is integrating Claude across its operations and giving all employees access to the software, according to Yahoo Finance.
For your portfolio, that means watching three signals over the next quarter. How Microsoft talks about its new non-exclusive OpenAI deal. Whether Anthropic's IPO timeline accelerates ahead of its rival's. And whether any major OpenAI researcher walks the other way for the first time in two years.
That is the part of the story that should keep Sam Altman awake. Losing a founding member to your biggest rival is bruising on its own. Losing him at the same time prediction markets, enterprise contracts and secondary valuations all start tilting the same direction is something different.
The AI race is no longer about who has the most compute. It is about who can attract the people who decide what to do with it. On that scoreboard, Anthropic just added a name that does not move often.
Related: Jim Cramer has a surprising take on Elon Musk's OpenAI loss
The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc.
This story was originally published May 20, 2026 at 5:37 PM.