Google DeepMind

Overview

Google DeepMind is Alphabet’s artificial intelligence research division — the 2023 merger of the London lab DeepMind with Google’s internal AI division, Google Brain — and the organization behind AlphaFold, the protein-structure-prediction system that won the 2024 Nobel Prize in Chemistry (Business Today). The standard overviews still describe it as “the lab Demis Hassabis runs.” They are out of date: in 2026 DeepMind disbanded its Nobel-winning AlphaFold team to pour resources into Gemini, and on August 5 Hassabis gave up day-to-day leadership to become the division’s Chair and Alphabet’s Chief Scientist (Scientific American; 9to5Google).

2026: the AlphaFold team folds

After eight years, more than 200 million predictions, and a Nobel Prize for two of its creators, the dedicated AlphaFold team “has itself folded,” as Scientific American put it in August 2026, following the Financial Times’ late-July report of the dissolution (Scientific American). Google reassigned key members and original paper authors to Gemini-related projects and other scientific research; a few left the company entirely (Engadget).

The most prominent departure was AlphaFold2 co-creator John Jumper, who announced in June 2026 that he was leaving for Anthropic — with a couple of colleagues following him there — while other former staff moved to Alphabet’s drug-discovery spin-off Isomorphic Labs (Engadget). A DeepMind spokesperson told Scientific American that many of the moves had happened more than a year earlier, and stressed that the program’s public database and prediction server remain available (Scientific American).

The breakup, per Engadget’s reporting, reflects Google’s decision to concentrate resources on Gemini. Pushmeet Kohli, DeepMind’s VP of research, framed it as evolution: “Our strategy over the last nine years has been to focus on grand challenges… a concrete goal every project is focused on. The strategy has evolved” (Engadget). A DeepMind spokesperson added: “We are incredibly proud of our scientific heritage and the global impact of AlphaFold, which inspired the launch of Isomorphic Labs with the mission to reimagine drug discovery” (Engadget). Scientific American also noted reported internal friction — Hassabis’s decision to make AlphaFold widely available at no cost brought prestige but little direct revenue inside Google (Scientific American).

Scientists outside the company were not alarmed. CASP co-founder John Moult, who in 2020 declared AlphaFold had “largely solved” structure prediction, called the move “not surprising”: “They decided that this was a good problem where they could show whether they succeeded or not in a clean way — not only show the world but genuinely show themselves” (Scientific American).

The August 2026 handover

Days after the AlphaFold news broke, Google announced the leadership change. Hassabis handed over “day-to-day operational responsibilities” to become Chair of Google DeepMind and Chief Scientist of Alphabet, a move he described as giving him “the time and space to focus on the big picture” of AGI (9to5Google). Day-to-day leadership went to former DeepMind CTO Koray Kavukcuoglu, now senior vice president reporting to Google CEO Sundar Pichai, overseeing “Gemini model development, Frontier AI research, and the Gemini app and developer teams” (9to5Google). Pichai’s endorsement was emphatic: “Koray has been at DeepMind since its early days, and over his 13 years there, he has started our deep learning team and led the way on breakthroughs like WaveNet and DQN” (9to5Google). Hassabis remains CEO of Isomorphic Labs (9to5Google).

From DeepMind to Google DeepMind

DeepMind was co-founded in 2010 by Demis Hassabis alongside Shane Legg and Mustafa Suleyman, and acquired by Google for $400 million in 2014 (Business Today). In April 2023 it was merged with Google AI’s Google Brain division, forming the new entity Google DeepMind — a consolidation aimed at concentrating Google’s AI research firepower (Business Today).

Its defining scientific achievement was AlphaFold. DeepMind began developing the system in 2018 to solve the “protein folding problem” — predicting a protein’s 3D structure from its amino acid sequence — a challenge scientists had worked on since the 1970s, having identified only about 170,000 protein structures over five decades using tools like X-ray crystallography and nuclear magnetic resonance (Scientific American; Engadget). In 2020, AlphaFold2 dominated CASP, the field’s blind test; in 2021 Nature published the methodology papers and DeepMind released predictions for the entire human proteome, launching the AlphaFold Protein Structure Database with free access to over 200 million predictions (Engadget). By 2022 the tool had predicted nearly all known protein structures and had been accessed by more than 2 million researchers in 190 countries (Business Today). In October 2024, Hassabis and Jumper shared half of the Nobel Prize in Chemistry for AlphaFold2 (the other half went to David Baker for computational protein design), for fulfilling what the Nobel Committee called “a 50-year-old dream” (Business Today).

What this profile does not claim

This profile does not give a precise count of AlphaFold team departures — the figure seen in secondary reports was not verified in a fully-read source; only the reassignments and named departures stated here are claimed. It does not present the reported internal friction over AlphaFold’s commercial returns as established fact; it is attributed to reporting. It does not predict whether the public AlphaFold database and prediction server will remain available long-term, only that DeepMind said they remain available. The Gemini product roadmap and the effects of the August 2026 reorganization are stated only as announced.