ORGANIZATION
DeepMind
Founded in London in 2010 by Demis Hassabis, Shane Legg, and Mustafa Suleyman. Acquired by Google in 2014. Known for deep reinforcement learning on Atari games (2013), AlphaGo in Go (2016), and the protein-structure breakthrough AlphaFold (2018 and 2020). Merged with Google Brain in 2023 to form Google DeepMind.
Organization
- Founded
- 2010
- Status
- Active
- Active for
- 16 years
- Appearances
- 02
- Name
- ENDeepMindJADeepMind
DeepMind
DeepMind is an AI research laboratory founded in London in September 2010 by Demis Hassabis, Shane Legg, and Mustafa Suleyman. Acquired by Google for about GBP 400 million in 2014, it was merged with Google Brain in April 2023 to form Google DeepMind. Starting from reinforcement learning, it has led the world in game-playing AI, protein-structure prediction, and general-purpose LLMs (Gemini).
History
Hassabis brought a PhD in neuroscience and a stint at Lionhead Studios; Legg, a logical theory of AGI; Suleyman, social-entrepreneurship experience. The founding theme was "general intelligence informed by neuroscience". An early Series A came from Peter Thiel, Founders Fund, and Horizons Ventures. The December 2013 NIPS paper "Playing Atari with Deep Reinforcement Learning" introduced the Deep Q-Network (DQN), and Google announced the acquisition in January 2014. Press reporting at the time noted that the deal included a commitment to an independent ethics board.
Nature ran a 2015 cover paper on a DQN that reached human-level performance on Atari. In March 2016 AlphaGo beat the world-class Go player Lee Sedol 4-1. AlphaZero unified Go, shogi, and chess via self-play (2017); AlphaFold (2018 and 2020) operationalised protein-structure prediction, culminating in the October 2024 Nobel Prize in Chemistry shared by Hassabis, John Jumper, and David Baker. From 2022 onward DeepMind ran LLM development in parallel with Google Brain — Chinchilla (compute-optimal scaling laws), Sparrow, Gemini. The April 2023 merger with Brain created Google DeepMind; Gemini 1.0 (December 2023), Gemini 1.5 (May 2024), Gemini 2.5 (March 2025), and Gemini 3 (November 2025) put the lab back at the frontier.
Products and Services
| Area | Representative projects | Notes |
|---|---|---|
| Game AI | DQN, AlphaGo, AlphaZero, MuZero, StarCraft II AlphaStar | Hallmark reinforcement-learning results |
| Science AI | AlphaFold 2 / 3, GNoME (materials), AlphaGenome, AlphaProteo, AlphaQubit | Proteins, materials, drug discovery, quantum error correction |
| General AI | Gemini 1-3, Gemini Nano, Imagen, Veo, NotebookLM | Google's flagship AI products |
| Robotics | RT-2, Gemini Robotics, ALOHA Unleashed | Vision-Language-Action models |
| Maths and reasoning | AlphaProof, AlphaGeometry, FunSearch | Silver-medal-level on the International Mathematical Olympiad |
| Drug discovery | Isomorphic Labs (spun out 2021) | Partnerships with Eli Lilly and Novartis; clinical trials in 2026 |
The AlphaFold protein-structure database now contains over 200 million predicted structures — essentially every known protein — and is used by more than 3 million researchers in over 190 countries.
Key Events
- September 2010: Founded in London.
- January 2014: Google acquires DeepMind for about GBP 400 m.
- February 2015: Nature paper — a single DQN plays 49 Atari games at human level.
- March 2016: AlphaGo defeats Lee Sedol.
- October 2017: AlphaGo Zero learns from scratch; AlphaZero generalises to chess and shogi in December.
- July 2018: Internal dispute over military pledges; Suleyman later leaves to co-found Inflection AI (2022), then joins Microsoft AI (2024).
- November 2020: AlphaFold 2 effectively wins CASP14.
- July 2021: AlphaFold 2 source and database released; Isomorphic Labs spun out.
- April 2023: Merged with Google Brain into Google DeepMind.
- May 2024: AlphaFold 3 released (predictions for protein-ligand complexes).
- October 2024: Hassabis and Jumper share the Nobel Prize in Chemistry.
- July 2025: AlphaProof reportedly achieves gold-medal-equivalent performance on the International Mathematical Olympiad.
- November 2025: Gemini 3 released; widely regarded as overtaking OpenAI and Anthropic on major benchmarks.
- February 2026: Isomorphic Labs unveils IsoDDE, which doubles AlphaFold 3's accuracy for drug-candidate generation.
Key Metrics
- Staff: estimated 3,000-5,000 (75 at the time of acquisition)
- AlphaFold database: over 3 million users across 190+ countries
- Citations: AlphaFold 2 papers receive over 40,000 citations per year as of 2025
- Gemini App MAU: roughly 750 m for Google's combined AI surface at end-2025
- Parent infrastructure: Alphabet's 2026 capex guidance is on the order of US$180 bn
Influence and Criticism
DeepMind established the three-layer template — games, science, general — that has come to define modern AI research. AlphaGo persuaded the wider public that AI could match human strategic intuition; AlphaFold demonstrated the "AI for Science" genre. Both successes, however, occurred inside a commercial parent, and the timing and scope of open releases have repeatedly been contested. AlphaFold 3 was initially published without code, drawing academic criticism before a research-use licence was rolled out.
After the merger, the independent ethics-board commitments reported at the time of acquisition appear to have softened, and former staff have described a loss of research independence. The April 2023 departure of Geoffrey Hinton to speak openly about AI risk, the exit of co-founder Suleyman, and turnover among lead researchers including David Silver and Lila Ibrahim have all been visible. Hassabis has framed the Brain merger as an attempt to recover "startup pace".
On military use, the 2014 acquisition was widely reported to include a no-military-or-intelligence undertaking. In 2024 Google revised its overall AI policy in a direction that loosened blanket military restrictions, drawing renewed internal protest. The biosecurity risk that AlphaFold structures might enable weapons design has been examined empirically in red-team studies since 2023.
Appearances
- March 15, 2016AlphaGo Defeats Lee SedolDeepMind's AlphaGo won its five-game series against Lee Sedol, one of the world's top Go players, 4–1. The combinatorial explosion of Go had long been thought to put a machine victory at least a decade away; the combination of Monte Carlo tree search and deep reinforcement learning broke that wall. Move 37 in Game 2—'the divine move'—introduced a placement so unlike anything a human player would have chosen that the professional Go community discussed it for weeks.
- October 8–9, 2024Nobel Prizes for AI — Physics and Chemistry in the Same WeekOn 8 October, the Nobel Prize in Physics was awarded to John Hopfield (the Hopfield network) and Geoffrey Hinton (the Boltzmann machine and deep learning). The next day, the Chemistry prize went to David Baker (computational protein design) and Demis Hassabis and John Jumper (DeepMind's AlphaFold2 for protein structure prediction). Both basic-science prizes going to AI work in the same week was unprecedented—a recognition of machine learning as an established scientific field. Hinton himself remarked that AlphaFold lay 'on a direct line' from his work.