DeepMind Dismantled the Team That Won It a Nobel Prize

Google DeepMind no longer has a dedicated AlphaFold team. The group behind the protein structure prediction work that earned a share of the 2024 Nobel Prize in Chemistry has been broken up, with members reassigned internally and several key authors gone to Anthropic.

DeepMind has confirmed the moves and frames them as a deliberate change in strategy rather than a retreat from scientific work.

Who left and where they went

Nobel laureate John Jumper departed for Anthropic in June 2026 after nearly nine years at DeepMind. Two AlphaFold contributors, Jonas Adler and Alexander Pritzel, followed him there.

The wider pattern is more striking than any individual exit. Most authors of the original AlphaFold paper have been reassigned over the past year, and close to a quarter of the full-time Google DeepMind authors have left the company altogether.

Those who stayed are working on Gemini-focused projects, enzyme design, nuclear fusion, genomics, and drug discovery at Isomorphic Labs, Alphabet's separate therapeutics company.

The technology itself is not going anywhere

This is worth stating plainly, because "team disbanded" reads worse than the situation is for anyone who depends on the tool. AlphaFold is not being discontinued. The public database still hosts more than 200 million predicted structures, the AlphaFold Server is running, and AlphaFold 3 remains available for academic use.

What has ended is the standing team whose job was to advance that specific line of research as a distinct effort.

What the reorganisation says about lab priorities

Two readings, and both are probably partly true.

The charitable one is DeepMind's own: AlphaFold solved the problem it was built for. Maintaining a dedicated team around a solved problem is not obviously the best use of some of the strongest researchers in the field, and redistributing them into fusion, genomics and enzyme design spreads that capability across more open questions.

The less comfortable one is about gravity. Nearly every large lab is now organising around a single frontier model line, and Gemini is where DeepMind's resources concentrate. Specialised scientific programmes have to justify themselves against that pull, and Nobel-winning work turns out not to be automatically exempt.

The departures to Anthropic add a competitive dimension. Talent moving between labs is unremarkable in this market; a Nobel laureate and two co-authors moving together is a more pointed signal about where scientifically motivated researchers currently expect to do their best work.

Why this matters outside the research world

For most businesses, the direct impact is zero. Protein folding is not on your roadmap.

The transferable lesson concerns vendor dependency. AlphaFold is safe because it is a mature system with a public database and an existing user base. Plenty of impressive lab demos are not in that position. When a research team is reorganised, the projects that quietly stop being developed are the ones that never reached product status.

If a supplier's pitch leans on a research capability rather than a shipped, supported product, the durability of that capability is a fair question to ask. Who maintains it, what its release history looks like, and what happens if the team behind it is redeployed. Those are ordinary procurement questions, and the last year has given plenty of reason to keep asking them.

Sources: The Decoder, Engadget, The Next Web, Fortune