Nobel Laureate Jumper and Two More Leave Google for Anthropic

John Jumper, who shared the 2024 Nobel Prize in Chemistry for AlphaFold — the AI model that solved protein structure prediction after 50 years of unsolved biology — has left Google DeepMind for Anthropic. He joins two other recently departed researchers: Jonas Adler, a specialist in AI-powered scientific computing, and Alexander Pritzel, who worked on AI systems training. All three were considered core contributors to Google's AI capabilities.

The departures extend a pattern that has accelerated in 2026. Earlier this month, Noam Shazeer — co-author of the 'Attention Is All You Need' paper that introduced the transformer architecture, and co-lead of Google's Gemini project — announced he was joining OpenAI. Google is not losing peripheral researchers; it's losing the people who built the systems it competes with.

The numbers tell the story

A SignalFire analysis quantified the directional asymmetry: DeepMind engineers switch to Anthropic eleven times more often than Anthropic engineers move to DeepMind. That kind of lopsided flow doesn't happen by chance or because of individual circumstances — it reflects a systemic difference in what one set of organizations can offer the other can't.

The primary driver appears to be equity. Both Anthropic and OpenAI are approaching initial public offerings, which makes their equity packages significantly more compelling than Google's current compensation structure. For a researcher who has spent years helping build technology that's reshaping the world, the prospect of meaningful financial upside tied to a company at an earlier growth stage is a different kind of offer than what a trillion-dollar incumbent can put on the table.

What this means for the models you'll use

Demis Hassabis, CEO of Google DeepMind, pushed back with a reasonable counterpoint: Google has more researchers across more disciplines than any other AI lab. The bench is deep. One departure, or even a cluster of departures, doesn't hollow out an organization of thousands. And talent flow in competitive industries runs in multiple directions — Google also recruits from competitors.

But there's a signal worth taking seriously beneath the noise. When the researcher who won a Nobel Prize for AI's most celebrated scientific achievement decides his next chapter belongs at Anthropic, that's a considered bet on where ambitious, fundamental research will happen next. The same is true of the Transformer co-author choosing OpenAI. These aren't people optimizing for job security — they're making long bets about where the most consequential work will be done. For anyone who watches AI closely, those bets are worth tracking.