Andrej Karpathy
Slovak-Canadian AI researcher Andrej Karpathy has worked at OpenAI, Tesla, and now Anthropic. YT/ No Priors: AI, Machine Learning, Tech, & Startups

Anthropic, the AI company closing in on a $1 trillion (£740 billion) private valuation, landed one of the most sought-after researchers in artificial intelligence on 19 May when OpenAI co-founder Andrej Karpathy announced he had joined the rival lab.

'I think the next few years at the frontier of LLMs will be especially formative. I am very excited to join the team here and get back to R&D,' Karpathy wrote on X. He added that he remains 'deeply passionate about education' and intends to return to that work eventually.

Anthropic confirmed to TechCrunch that Karpathy began work this week under Nick Joseph, who leads the pre-training division. That team handles massive, resource-intensive training runs that form the foundation of what Claude knows and can do. Karpathy's specific brief is to establish a new unit that will use Claude to accelerate the research behind those training runs—betting on AI-assisted science rather than raw computing power to keep pace with OpenAI and Google.

Who Is Andrej Karpathy? The Man Who Coined 'Vibe Coding'

Karpathy, 39, was born in Bratislava and raised in Toronto. He earned a PhD in computer science at Stanford under Fei-Fei Li, then helped found OpenAI in 2015 as a research scientist specialising in deep learning and computer vision.

Elon Musk recruited him in 2017 to run artificial intelligence at Tesla. There, Karpathy built and oversaw the vision systems powering Autopilot and Full Self-Driving. He stepped down in 2022, returned briefly to OpenAI for about a year, then left once more in early 2024. He described that departure as a personal decision, not a reaction to any fallout inside the company.

He then went on to build Eureka Labs, an AI education startup, and quickly became one of the field's most visible public voices. His YouTube lecture series, Neural Networks: Zero to Hero, draws millions of views. His following on X sits close to two million.

But the moment that cemented his cultural footprint was in February 2025, when he posted about a 'new kind of coding' in which developers describe what they want in plain language and let the AI write the programme. He called it 'vibe coding'. Collins Dictionary later named it Word of the Year. The AI model he referenced in that post was Anthropic's Claude Sonnet.

Anthropic's Revenue Has Tripled Since December

The timing of the hire is no coincidence. Anthropic's annualised revenue run rate surged to $30 billion (£22.2 billion) by April 2026, up from roughly $9 billion (£6.66 billion) at the end of 2025, according to figures shared by CEO Dario Amodei.

Much of that growth has been fuelled by Claude Code, the company's agentic coding tool, which went public in May 2025 and hit $2.5 billion (£1.85 billion) in run-rate revenue by February 2026.

Bloomberg has reported that the firm is in early discussions to raise at least $30 billion (£22.2 billion) in fresh capital at a pre-money valuation above $900 billion (£666 billion). If that round closes, it would push Anthropic past OpenAI, which was last valued at $852 billion (£630.5 billion) in March.

A Growing Roster of Senior Defections

Karpathy's move came just one day after the Musk v. Altman trial wrapped up on 18 May. A jury and federal judge sided with OpenAI CEO Sam Altman. Karpathy's contributions at both OpenAI and Tesla featured prominently in the proceedings, CNBC noted.

Anthropic was founded in 2021 by Dario and Daniela Amodei, both former OpenAI research executives. Karpathy now joins a growing list of talent that crossed from one lab to the other.

He is not the only notable arrival. Ross Nordeen, a founding member of Musk's xAI and a former Tesla employee, signed on with Anthropic earlier in May. The company separately announced that cybersecurity veteran Chris Rohlf has joined its frontier red team. Rohlf spent six years at Meta and previously worked on Yahoo's security unit, widely known in the industry as 'The Paranoids'.

Funding rounds and GPU supply chains tend to dominate headlines in the AI sector. But the quiet migration of senior researchers like Karpathy tells its own story about which companies the people closest to the technology believe will shape what comes next.