The team that won a Nobel Prize for cracking protein folding has been dismantled. Google DeepMind confirmed to the Financial Times that it reassigned most of the researchers behind AlphaFold over the past year, redirecting them toward projects built around the Gemini large language model, enzyme design, nuclear fusion, and genomics. Others moved to Isomorphic Labs, the Alphabet drug discovery spinout. But the real shock is the exodus: nearly a quarter of the full-time Google DeepMind authors from the original AlphaFold papers have left the company entirely.
This isn’t just a routine reorg. It’s a wholesale abandonment of the strategy that made DeepMind famous. For nearly a decade, the lab organized itself around ‘grand challenges’ — concrete, long-term scientific problems tackled by dedicated teams. Pushmeet Kohli, VP of research, told the FT plainly: ‘The strategy has evolved.’ The new playbook? Building Gemini-powered AI assistants for scientists and competing head-to-head with OpenAI and Anthropic on frontier AI agents. Pure science just got sidelined for product.
The talent drain hits the inner circle. John Jumper, the Nobel laureate and VP who led AlphaFold, initially moved to Google’s internal ‘Code Strike’ team to improve AI coding tools — a direct shot at catching Anthropic’s Claude. Last month, he announced he’s leaving for Anthropic entirely, taking fellow AlphaFold researchers Jonas Adler and Alexander Pritzel with him. One unnamed DeepMind employee told the FT the trio were ‘instrumental, important, core members’ and their departures sparked surprise internally. That’s corporate speak for panic.
What’s left is a lab chasing rivals rather than defining its own lane. DeepMind says it’s ‘incredibly proud’ of AlphaFold’s legacy, but the actions tell a different story. Anthropic launched Claude Science for biology earlier this year. Isomorphic Labs has the pharma partnerships. DeepMind is betting Gemini can automate scientific discovery — but it’s doing so without the team that proved it actually could. The question nobody’s answering: if the people who cracked protein folding aren’t excited about where the lab is going, why should anyone else be?
