Andrej Karpathy, a co-founder of OpenAI and former Tesla AI chief, joined Anthropic this week to lead a new team focused on using Claude to accelerate the most compute-intensive phase of building frontier AI models. The hire underscores how Anthropic plans to stay competitive with OpenAI and Google by investing in research efficiency rather than simply scaling up computing power.

Who Is Andrej Karpathy and Why Does His Move Matter?

Karpathy is one of the few researchers who can bridge the gap between large language model (LLM) theory and large-scale training practice. At OpenAI, he focused on deep learning and computer vision before leaving in 2017 to join Tesla, where he led the Full Self-Driving and Autopilot programs. He returned to OpenAI for one year before departing in 2024 to start Eureka Labs, an education-focused startup.

His move to Anthropic signals a critical moment in frontier AI development. Karpathy announced the decision on X, stating that "the next few years at the frontier of LLMs will be especially formative" and expressing excitement about returning to research and development. He started at Anthropic on May 19, 2026, reporting to pretraining lead Nick Joseph.

Karpathy

What Will Karpathy Do at Anthropic?

Karpathy will build a team using Claude to speed up pretraining, the most compute-intensive and expensive phase of frontier model development. Pretraining is where models learn their core capabilities by processing vast amounts of text data. Even small efficiency gains at this stage translate to tens of millions of dollars in compute savings.

The focus on AI-assisted research rather than pure compute expansion reveals Anthropic's strategic bet: research velocity, not just raw computing power, will determine which labs build the next generation of frontier models. If Claude can accelerate its own pretraining research, Anthropic gains a compounding advantage over competitors still relying on human-paced iteration.

How Anthropic Is Building Its Competitive Edge

• Talent Acquisition: Anthropic has been aggressive in recruiting top-tier technical talent, with Karpathy's hire representing a major win in the high-stakes competition for elite AI researchers.
• Safety and Security Investment: The company simultaneously brought on Chris Rohlf to its frontier red team, which stress-tests advanced AI models against severe threats. Rohlf brings 20 years of cybersecurity experience, including six years at Meta and prior work at Yahoo's cybersecurity team known as The Paranoids.
• Infrastructure Control: Anthropic recently acquired Stainless for over 300 million dollars, pulling the software development kit (SDK) tool from OpenAI and Google to own more of the developer experience.

"I've joined Anthropic. 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," said Andrej Karpathy.

Andrej Karpathy, AI Researcher and OpenAI Co-founder

Rohlf's background in offensive security and AI-driven cybersecurity research positions him to probe Claude for vulnerabilities at scale, underscoring Anthropic's continued investment in safety as models grow more capable.

What Are the Risks and Opportunities?

The risk for Anthropic is execution. Karpathy has a track record of departures, having left OpenAI twice and Tesla once, and his ability to scale a team inside Anthropic's organizational structure remains untested. However, the upside is substantial: if Claude can accelerate its own pretraining research, Anthropic gains a compounding advantage over labs still relying on human-paced iteration.

For Karpathy personally, the move back into frontier research suggests he sees the next few years as a critical inflection point in AI development. His work at Anthropic will likely focus on closing the loop between research and deployment at the scale where even small efficiency gains translate to significant cost reductions. Whether that bet pays off will depend on how quickly his team can turn Claude into a tool that makes itself better.

Karpathy has also maintained a strong educational presence, teaching an online course called "Neural Networks: Zero to Hero" and maintaining a YouTube channel with lectures on LLMs and AI. In a statement, he noted, "I remain deeply passionate about education and plan to resume my work on it in time".