Bloomberg reported that John Jumper, the Nobel-Prize-winning leader of AlphaFold and a long-time Google DeepMind researcher, is leaving Google DeepMind to join Anthropic. The move is a major signal for the LLM × AI4Science (AI for Science) field.
Jumper's background: he led the AlphaFold team that solved the protein structure prediction problem, winning the Nobel Prize in Chemistry in 2024. AlphaFold is widely considered the most successful AI-for-Science project to date, and Jumper is one of the most respected figures in the field.
The "LLM × AI4Science" angle: at Anthropic, Jumper is expected to lead a new "AI4Science" team that will apply LLM techniques to scientific discovery. The "LLM × AI4Science" combination is one of the most promising directions in AI — the LLM's reasoning capability combined with scientific domain knowledge could lead to breakthroughs in drug discovery, materials science, and climate modeling.
The implications for Google: losing Jumper is a significant blow to Google DeepMind's AI4Science ambitions. DeepMind has been the leader in AI4Science (AlphaFold, AlphaFold-Multimer, AlphaMissense), and the loss of its most prominent figure raises questions about the future of these efforts.
The bigger takeaway: "AI4Science" is becoming a major LLM battleground. The "AI for code, AI for chat" era is being augmented with "AI for science" — i.e., using LLMs to accelerate scientific discovery. Anthropic, OpenAI, and DeepMind are all investing heavily in this direction, and the talent war will shape who leads.