Andrej Karpathy, an artificial intelligence researcher who was a co-founder of OpenAI and later held a senior AI role at Tesla, is joining Anthropic, the company announced on Tuesday. Karpathy is set to begin work this week and will join Anthropic's pretraining team.
Anthropic described his new role as centering on pretraining - the phase in which its Claude models acquire foundational knowledge and capabilities. The company said Karpathy will establish and lead a team that uses Claude to accelerate pretraining research.
In a post on X about the move, Karpathy wrote that the coming years at the frontier of large language models will be "especially formative," and said he is excited to return to research and development. The announcement did not provide further operational details about the team he will assemble or specific research timelines.
The hire follows other high-profile additions to Anthropic's staff. Earlier this month Ross Nordeen, a founding member of xAI and a former Tesla employee, said he was joining Anthropic. The personnel news coincided with a separate development in Anthropic's compute strategy: the company struck a deal to rent compute capacity at xAI's Colossus 1 data center in Memphis, Tennessee, which is operated by SpaceX.
Karpathy's background includes helping to start OpenAI and then leaving to join Tesla in 2017. At Tesla he served as director of AI and led the computer vision team working on the Autopilot program. The announcement from Anthropic did not include additional comments from Karpathy beyond his X post or a detailed timeline for the team's research milestones.
Summary
Andrej Karpathy is joining Anthropic's pretraining team this week to lead efforts using Claude to speed pretraining research. His hire is part of a broader wave of talent recruiting at Anthropic and comes alongside the company's compute arrangements with SpaceX at the Colossus 1 facility in Memphis.
- Key points
- Karpathy will focus on pretraining work for Anthropic's Claude models, building a dedicated team to use Claude in pretraining research.
- Anthropic has continued to add prominent AI staff, including Ross Nordeen earlier this month, indicating intensified competition for talent.
- The company has secured compute capacity at xAI's Colossus 1 data center in Memphis, Tennessee, via a deal with SpaceX, highlighting the role of external compute resources.
- Risks and uncertainties
- Competition for top AI talent remains intense, which could affect recruitment costs and retention - relevant to technology and cloud computing sectors.
- Reliance on rented compute capacity introduces operational dependency on third-party data center resources - relevant to cloud/compute and infrastructure markets.
- The announcement does not provide specific timelines or measurable milestones for the new pretraining team, leaving outcomes and development pace uncertain for investors and partners in tech and AI research.