Nobel Laureate John Jumper Leaves DeepMind for Anthropic After Nearly Nine Years
John Jumper, who shared the 2024 Nobel Prize in Chemistry with Demis Hassabis for developing AlphaFold2, announced Friday he is leaving Google DeepMind to join Anthropic. “After nearly nine years, I have decided to leave DeepMind,” Jumper wrote on X.
Jumper was the lead implementation researcher behind AlphaFold2, the protein structure prediction system that solved a problem biologists had worked on for 50 years. His departure is the most significant loss DeepMind has absorbed this year — a direct line from the lab’s defining scientific achievement to a commercial competitor.
What Jumper Brings to Anthropic
Anthropic has been building quietly into scientific AI. Claude has been used across chemistry, biology, and drug discovery applications. Jumper’s arrival signals Anthropic is now competing seriously for frontier scientific talent, not just software engineers.
His technical profile is distinctive: a computational biologist who understands both the deep physics of molecular systems and the architectural decisions that make neural networks work at scientific scale. AlphaFold2 was not a brute-force model. It required architectural innovations, including novel attention mechanisms for biological sequences, that have since influenced model design across the field.
Anthropic has not disclosed what role Jumper will take. The company’s research leadership already includes several figures who came from fundamental research backgrounds rather than pure machine learning.
DeepMind’s Talent Drain
The departure follows a pattern at DeepMind that has accelerated since Google merged the lab with Google Brain into Google DeepMind in 2023. Several senior researchers have left for competitors or to start independent labs.
Jumper’s case stands out because he won a Nobel Prize while at DeepMind — recognition that typically anchors researchers to an institution. His departure suggests the commercial and scientific opportunities at frontier AI companies now outweigh the institutional prestige that once held top researchers at large technology labs.
DeepMind still has Hassabis and other AlphaFold contributors, but the co-lead of its most celebrated scientific project has moved to a direct competitor. The practical implications for DeepMind’s science roadmap are unclear.
AlphaFold and What Comes Next
AlphaFold2 solved protein structure prediction. AlphaFold3, released in 2024, extended the approach to DNA, RNA, and small molecules. The open question in structural biology is whether similar methods can design novel proteins and drugs from scratch, not just predict what already exists.
Anthropic’s interest in biology is documented. The company has published work on interpretability and alignment but also supports scientific applications through its API and research collaborations. Whether Jumper is joining a dedicated biology research team or contributing to core model architecture is not yet public.
His move is the clearest signal this year that the talent market in foundational scientific AI has consolidated around the frontier model companies. Anthropic is now competing for Nobel-level credentials, and winning them.