Google DeepMind Disbands Nobel Prize-Winning AlphaFold Team, Redirects Staff to Gemini

Google DeepMind has dismantled the team behind AlphaFold, the AI system that solved a 50-year-old scientific puzzle and won its creators a Nobel Prize. According to the Financial Times, most authors of the original AlphaFold papers have been reassigned over the past year. Nearly a quarter of the full-time authors still at the company when the shift began have since left entirely.

AlphaFold predicts the 3D structure of proteins from their amino acid sequences, a job that used to take scientists years of lab work with X-ray crystallography and nuclear magnetic resonance. DeepMind started building it in 2018. By 2020 it had effectively cracked the "protein folding problem." In 2021, Nature published the methodology along with structure predictions for the entire human proteome. DeepMind then opened the AlphaFold Protein Structure Database to researchers worldwide, offering more than 200 million protein structure predictions for free.

In 2024, DeepMind CEO Demis Hassabis and lead researcher John Jumper won the Nobel Prize in Chemistry for the work. Jumper started at DeepMind as a staff research scientist and rose to vice president and engineering fellow by the time the prize was announced.

Jumper announced in June that he was leaving DeepMind for Anthropic. A couple of his colleagues followed him out the door. DeepMind confirmed to the Financial Times that other staff members were moved internally onto Gemini-focused projects, while others were shifted to Alphabet's drug-discovery spinoff, Isomorphic Labs. Some researchers landed on new science projects, including enzyme design, nuclear fusion, and genomics.

Pushmeet Kohli, DeepMind's vice president of research and the person who built the lab's "AI for Science" strategy, told the Financial Times the company is walking away from a nine-year approach centered on "grand challenges... a concrete goal every project is focused on." His quote: "The strategy has evolved."

In other words: the scientific-moonshot model is out, and the general-purpose AI agent model is in. DeepMind is now competing head-on with OpenAI and Anthropic to build Gemini into an autonomous research and reasoning system, not just a chatbot.

The case for Google's move

There's a legitimate business argument here. Alphabet is spending tens of billions of dollars a year on AI infrastructure, and Gemini is the product actually generating consumer traffic, enterprise contracts, and competitive pressure from OpenAI and Anthropic. AlphaFold, while scientifically enormous, doesn't monetize the way a foundation model does. Isomorphic Labs, the drug-discovery spinout, is where DeepMind is betting the science-focused work can still turn a profit through pharma partnerships rather than open research.

Redeploying elite researchers toward the product that actually competes in the market Google is trying to win is a defensible allocation of talent, especially when those researchers' skills, like large-scale model training and evaluation, transfer directly to Gemini work. Nobel Prizes don't pay Alphabet's cloud bills.

The problem with that argument

AlphaFold wasn't a side project. It was arguably the single clearest example in the last decade of AI producing unambiguous, verifiable scientific value. Winning a Nobel Prize for it in 2024 and then hollowing out the team less than two years later sends a signal: even historically significant scientific work gets deprioritized the moment there's a hotter product to chase.

The departures make that concrete. John Jumper didn't get promoted within DeepMind's new priorities. He left for a direct competitor. So did colleagues who worked alongside him. Researchers were reassigned to adjacent work, but talent voted with its feet.

What Engadget and BigGo left out and got right

Engadget's coverage focused on the reassignments and Kohli's quote but didn't draw the connection between DeepMind's competitive positioning and why losing Jumper to Anthropic specifically matters, since Anthropic is one of the two companies DeepMind is now racing against. BigGo Finance's account fills that gap, framing the exits as a "brain drain" that puts DeepMind "in direct competition with OpenAI and Anthropic in the frontier AI agent race," and notes some remaining researchers were funneled into other science areas like nuclear fusion and genomics rather than purely Gemini work.

Neither source reports any statement from Hassabis himself addressing the departures directly. Alphabet has not disclosed how many researchers moved where, or what AlphaFold's ongoing maintenance and database support will look like going forward. The AlphaFold Protein Structure Database, still used by researchers globally, remains online, but its future development plans under a diminished team are unclear.