Andrej Karpathy joins Anthropic

Andrej Karpathy said on May 19 that he had joined Anthropic, adding one of the best-known researchers in artificial intelligence to the Claude maker’s ranks. Karpathy announced the move in a post on X, saying he was returning to research and development and believed “the next few years at the frontier of LLMs will be especially formative.” CNBC and Reuters reported that Anthropic said he would be part of its pretraining team, the group that works on the core training runs behind Claude models. The announcement drew wide attention across the AI industry and on social media. ### Who is Andrej Karpathy, and why does this hire stand out? Andrej Karpathy is a founding member of OpenAI and a former Tesla AI leader, according to Reuters and CNBC. His work has made him one of the most recognizable figures in the field, both through research roles and through public technical education. Stanford-trained Karpathy has moved between frontier labs and industry before. Reuters described him as a former Tesla AI executive and one of OpenAI’s founding members, while CNBC said he had previously been recruited from OpenAI to Tesla before later leaving OpenAI again in 2024. ### What exactly did Karpathy say he will do at Anthropic? Karpathy said in his May 19 X post that he had “joined Anthropic” and was “very excited to join the team here and get back to R&D.” Reuters quoted the same post as saying the coming years at the frontier of large language models would be “especially formative.” CNBC reported that Anthropic said Karpathy would be part of the pretraining team. That group helps Claude models acquire their core knowledge and capabilities, CNBC said, placing him in one of the most compute-intensive parts of model development. ### Why does pretraining matter inside a company like Anthropic? Anthropic’s own materials describe the company as an AI safety and research company building “reliable, interpretable, and steerable AI systems.” The company’s product pages and research posts show it is still investing heavily in frontier-model development, including Claude 4 and other recent releases. Claude 4 remains a central benchmark point for Anthropic’s model lineup. On Anthropic’s product page for Claude 4, the company said Opus 4 led on SWE-bench at 72.5% and Terminal-bench at 43.2%, underscoring why pretraining and core model capability remain a focus for the company’s research organization. ### How big was the public reaction to the announcement? Karpathy’s May 19 post drew heavy engagement on X, according to the public metrics cited in the source briefings. Those figures showed more than 146,000 likes, about 11,000 reposts and roughly 7,700 replies as of May 21. The scale of that response reflected Karpathy’s standing with developers and AI researchers. Reuters and CNBC both treated the announcement as a notable talent move in the competition among frontier-model companies. ### What else is Anthropic doing right now? Anthropic’s newsroom shows a string of May announcements that place the hire in a broader expansion push. On May 19, the company announced a strategic alliance with KPMG to integrate Claude across the consulting firm’s workforce of more than 276,000, according to Anthropic’s news page. Earlier in May, Anthropic announced a partnership with the Gates Foundation, a small-business Claude offering, and what it called a compute deal with SpaceX, according to the same newsroom listing. Those updates show the company pairing research hiring with product and enterprise rollouts. ### What comes next after the announcement? May 2026 is the clearest next checkpoint for this story because Anthropic is continuing to post product and enterprise updates on its newsroom page. Karpathy said he planned to focus on R&D for now, while adding in his X post that he remained passionate about education and planned to resume that work “in time.”