TL; DR
Andrej Karpathy, one of the original 11 co-founders of OpenAI, has joined Anthropic’s pre-prep team. He will set up a new team that uses Claude himself to speed up the most expensive phase of frontier model development.
Andrej Karpathy, one of the original co-founders of OpenAI and among the world’s most prominent AI researchers, announced on Monday he joined Anthropic. The move is a significant talent coup for the Claude maker, which is racing to stay on the cutting edge of big-language model development.
Andrej Karpathy X update, source: X
Karpathy joins Anthropic Nick Joseph’s pre-workout team, where he will build a new group focused on a compelling recursive goal: using Claude himself to accelerate pre-workout research. The pre-training, massively computationally demanding phase that gives the boundary model its basic knowledge and capabilities is the single most expensive part of building systems like Claude. Finding ways to make this process faster and more efficient could reshape the economics of the entire AI industry.
In an X post that has garnered 13.6 million views, Karpathy wrote that “the next few years at the frontier of LLMs will be particularly formative.” He added that he is “very passionate about education” and plans to resume the job in due course.
Recruiting spans a career arc that touches nearly every major turning point in modern AI. Karpathy earned his PhD at Stanford from Fei-Fei Li, the computer scientist behind ImageNet, focusing on deep learning and computer vision. He was among 11 people Co-founder of OpenAI Here in 2015, he worked on deep learning research before leaving to join Tesla in 2017 as director of artificial intelligence.
At Tesla, Karpathy led the computer vision teams behind the Full Self-Driving and Autopilot programs that support the electric carmaker’s ambitions for autonomous vehicles. He left in July 2022, returned to OpenAI for about a year, then left again to found Eureka Labs, a startup that filed in 2024. AI assistants in education. Eureka Labs is now on hold as Karpathy throws his weight behind Anthropic.
The timing is remarkable. Anthropic has emerged as a magnet for top technical talent at a time when its arch-rival OpenAI has. experienced a number of high-profile departures. Over the past two years, OpenAI has lost more than a dozen senior managers and researchers, including CTO Mira Murati, reinforcement learning pioneer John Schulman, and most recently three managers who left in one day in April 2026.
For Anthropic, the Karpathy landing shows the company can attract top talent as it expands both its research and commercial operations. The firm, led by CEO Dario Amodei, has attracted investor interest with a valuation of around $800 billion and is reportedly exploring an IPO that could come as late as 2026.
Karpathy’s new role also highlights a broader trend in frontier AI: using existing models to improve the next generation. If Claude can speed up his pre-training pipeline meaningfully, it will be a practical demonstration of recursive self-improvement, a capability the AI security community has long watched closely. Whether this prospect excites or frustrates observers may depend on how much they believe it security-minded culture Anthropic cultivated since its inception.
For now, Karpathy seems exactly where he wants to be: back in the lab, building models on the frontier.
