John Jumper, AlphaFold's Nobel-laureate co-creator, leaves Google DeepMind for Anthropic

John Jumper, the AlphaFold co-creator who shared the 2024 Nobel Prize in Chemistry for protein-structure prediction, told followers on 2026-06-19 that he is leaving Google DeepMind after nearly nine years to join Anthropic. The exit pulls a sitting VP and Engineering Fellow out of DeepMind’s life-sciences bench in the same news cycle that took Gemini co-lead Noam Shazeer to OpenAI, and it lands while Anthropic is preparing a public science event for 2026-06-30 and working through a confidential S-1 review.

What he said and where he is going

The post on X is short: “After nearly nine years, I have decided to leave Google DeepMind and join Anthropic,” followed by a thanks to DeepMind colleagues and a line saying he is excited about the discoveries still to come at his former employer. Anthropic has not publicly named his role or research charter. Jumper led the AlphaFold team that predicted more than 200 million protein structures, the work the Nobel Committee cited in 2024, and he most recently held a VP and Engineering Fellow seat that put him on DeepMind’s coding-team agenda as well as the life-sciences track. Google DeepMind’s statement was brief: thanks for the contributions, wishes for the next chapter.

Where this lands in the market

The cumulative bench shift is the story. Within roughly a month, the Gemini co-lead is at OpenAI, AlphaGo researcher David Silver has left to start his own reinforcement-learning lab, and now AlphaFold’s co-creator is at Anthropic. Three of DeepMind’s most heavily cited researchers from the decade are now outside the building. For developers tracking the next model cycle, the practical question is where the loss will show first. Anthropic’s recent product push has emphasized language-model agents (Claude Fable 5 and Mythos 5, the managed-agent integrations with Cloudflare and Modal). Bringing Jumper in positions the company to point a domain-fluent researcher at scientific-reasoning tooling that would compete directly with OpenAI’s GPT-Rosalind line and with the AlphaFold-adjacent surfaces that Google has historically owned. The hire is happening while Anthropic is also working under a US export-control restriction on its Mythos-class line and inside an S-1 process; the public recruiting pitch leans on research independence rather than scale of distribution.

What’s worth watching

1. A named Anthropic life-sciences product line. Jumper joining without a stated research charter is the most editorially interesting signal in the announcement. Whether Anthropic stands up a named life-sciences model (a Rosalind equivalent), or folds the work into a broader scientific-reasoning track inside the Mythos research program, will tell biology and drug-discovery teams where to point procurement asks in 2027.
2. Gemini 3.5 Pro launch credibility. Insider reports have suggested Gemini 3.5 Pro will not match the latest Anthropic and OpenAI lines on benchmarks. Losing the AlphaFold lead inside the launch window gives Google a harder narrative to argue, especially on the scientific-reasoning surfaces Gemini has historically led on. Watch whether the launch deck names new science leads to fill the rhetorical gap.
3. Anthropic’s S-1 governance section. If Jumper surfaces as a named officer in the eventual amendment, the public-market read is that Anthropic is treating science-product leadership as a board-level disclosure rather than a back-of-the-org-chart line. That reframing changes how analysts price the company against OpenAI’s enterprise-revenue story.

Sources cited

1. CNBC: John Jumper to leave Google DeepMind for Anthropic www.cnbc.com
2. Bloomberg: Nobel Winner John Jumper to Leave Google DeepMind for Anthropic www.bloomberg.com
3. The Decoder: Google DeepMind loses another top AI researcher as Nobel laureate John Jumper leaves for Anthropic the-decoder.com
4. John Jumper announcement post x.com