Andrej Karpathy joins Anthropic research
- Andrej Karpathy joined Anthropic on May 19, 2026, saying he would return to research and development on the company’s pretraining team. - Anthropic said Karpathy will build a team using Claude to accelerate pretraining research, the model-building stage that gives systems core knowledge. - Karpathy said he starts this week; Anthropic’s pretraining work is led by Nick Joseph, according to company statements.
Andrej Karpathy has joined Anthropic, adding one of the best-known researchers in artificial intelligence to the company’s pretraining effort as competition for senior technical talent intensifies. Karpathy announced the move on May 19 in a post on X, writing that “the next few years at the frontier of LLMs will be especially formative” and that he was “very excited” to get back to research and development. Anthropic said he starts this week and will build a team focused on using Claude to accelerate pretraining research. The move places a former OpenAI co-founder and ex-Tesla AI director inside one of OpenAI’s closest rivals. ### What exactly is Karpathy going to do at Anthropic? Anthropic said Karpathy will join its pretraining team, the group that works on the stage of model development where systems acquire core knowledge and capabilities. The company told multiple outlets that he will build a team focused on using Claude to speed up pretraining research. (cnbc.com) Karpathy said in his May 19 post that he was joining Anthropic because the coming years in large language models would be “especially formative.” Observer reported that he will help lead pretraining efforts, while CNBC said Anthropic described the role as part of the team responsible for Claude’s foundational training work. ### Why is this hire drawing attention beyond one personnel move? (cnbc.com) Anthropic has been adding senior researchers and infrastructure talent as it expands Claude and pushes deeper into frontier-model development. CNBC reported that Ross Nordeen, a founding member of xAI and former Tesla employee, also joined Anthropic earlier in May. Observer separately described Karpathy’s arrival as part of a broader rivalry between Anthropic and OpenAI for high-profile researchers. (observer.com) TechRepublic and Observer both framed the hire as a boost to Anthropic’s pretraining bench. That framing rests partly on Karpathy’s background: he helped found OpenAI, worked on early GPT efforts, and later led Tesla’s computer vision work for Autopilot. ### Why does Karpathy’s background matter here? Karpathy left OpenAI for Tesla in 2017, where he served as director of AI and led the computer vision team behind Tesla Autopilot, CNBC said. (cnbc.com) Observer reported that after five years at Tesla he returned to OpenAI, then left again in 2024 to launch Eureka Labs, an AI education startup. (techrepublic.com) Elon Musk’s past comments about Karpathy resurfaced in several reports this week. CNBC and Observer both cited court exhibits from Musk’s recent lawsuit against OpenAI in which Musk described Karpathy as “arguably the #2 guy in the world in computer vision,” behind OpenAI co-founder Ilya Sutskever. ### How does this fit into Anthropic’s recent momentum? (cnbc.com) Anthropic has paired recruiting moves with rapid business growth. CNBC reported on May 20 that the company was projected to reach $10.9 billion in revenue during the second quarter and was on track for its first profitable quarter, citing a source familiar with the matter. Reuters also reported that Anthropic was nearing its first quarterly operating profit and had agreed to pay SpaceX $1.25 billion a month for computing power. (cnbc.com) Those figures do not explain Karpathy’s hiring on their own, but they place the move inside a period when Anthropic is spending heavily on compute and staffing while scaling Claude. Observer noted that other former OpenAI figures, including Jan Leike and, briefly, John Schulman, have also moved to Anthropic. ### What happens next for Karpathy and Anthropic? (cnbc.com) Karpathy said he starts this week, and Anthropic said his initial task is to build a team around Claude-assisted pretraining research. Yahoo’s coverage, citing the company, said the broader pretraining team is led by Nicholas Joseph, an early Anthropic hire and former OpenAI employee. (observer.com) Karpathy also said he remains committed to education work and plans to return to it over time. For now, the next visible milestone is likely to be whether Anthropic names additional researchers joining the new group or discusses the effort in future Claude research updates. (observer.com) (cnbc.com)