OpenAI co-founder Andrej Karpathy joins Anthropic to lead AI research unit

Anthropic's annualised revenue has tripled to $30 billion since December and now it has pulled off the kind of hire that signals a company playing to win. 

Andrej Karpathy, one of the most recognised names in AI research, on Tuesday announced he had joined the Claude maker, ending months of public silence about his next move.

Who is Andrej Karpathy?

Karpathy, 39, co-founded OpenAI in 2015, right after he finished his PhD at Stanford, under the AI pioneer Fei-Fei Li. Elon Musk recruited him to Tesla in 2017, and there he built the computer vision machinery that powers Autopilot, plus Full Self-Driving, before he stepped down in 2022.

He came back to OpenAI for a short while, sort of in between things then left again in early 2024 to start Eureka Labs, which is basically an AI education startup.

His public presence is unmatched by any other active scientist. His YouTube series "Neural Networks: Zero to Hero" has been viewed millions of times, and his followers on X number around two million.

The term "vibe coding" was first coined by Karpathy in February 2025, when he described programming software through communicating with an AI program in simple terms, and Collins Dictionary chose it as Word of the Year. This was done using Anthropic's Claude Sonnet model.

TechCrunch verified with Anthropic that Karpathy works under the supervision of Nick Joseph, who oversees the pre-training department in charge of running the foundational training runs to shape Claude's knowledge and reasoning capabilities.

Karpathy's mission is to develop a new team that employs Claude to speed up the scientific research involved in the training runs.

The idea here is that AI-assisted science would enable Anthropic to compete with OpenAI and Google not through overspending on compute resources. "I think the next few years at the frontier of LLMs will be especially formative," Karpathy wrote on X when announcing his role.

Karpathy comes after Ross Nordeen, one of the founding members of Musk's xAI and a former Tesla engineer, who joined Anthropic last month. The company also hired cybersecurity expert Chris Rohlf, formerly with six years at Meta, for its frontier red team.