Skip to content
Longterm Wiki

Google DeepMind

Frontier AI Lab
Founded Sep 2010 (15 years old)HQ: Londondeepmind.google

Also known as: DeepMind, DeepMind Technologies, Google DeepMind

Entity
Wiki
About
Business
Policy & Governance
Output & Research
Data

Google DeepMind was formed in April 2023 from the merger of DeepMind and Google Brain, uniting Google's two major AI research organizations. The combined entity represents one of the world's most formidable AI research labs, with landmark achievements including AlphaGo (defeating world champions at Go), AlphaFold (solving protein folding), and G...

Headcount
6,000
as of Jun 2025
AI Models
6

Key Metrics

Headcount

6KJun 2025
Headcount chart. Employees: 1.6K in 2022 to 6K in 2025.01.7K3.5K5.2K6.9K20222025

Facts

13
Financial
Headcount6,000
Internal Revenue$1.7B
Safety & Research
Safety Researchers120
Model
Context Window2 million
Other
Parent OrganizationAlphabet Inc.
General
Websitehttps://deepmind.google/
Organization
HeadquartersLondon
Legal Structuresubsidiary company
Founded DateSep 2010
CountryUnited Kingdom
People
Founder (text)Demis Hassabis
Biographical
Wikipediahttps://en.wikipedia.org/wiki/Google_DeepMind

Other Data

Entity Events
14 entries
TitleDateEventTypeDescriptionSignificance
Gemini 2.0 launched2024-12launchEnhanced agentic capabilities; integrated across Google products.moderate
Gemini 1.5 launched2024-02launch2M-token context window; long-context leadership.moderate
Nobel Prize in Chemistry awarded2024milestoneAwarded to Demis Hassabis and John Jumper of DeepMind (shared with David Baker of University of Washington for independent protein-design work) for AlphaFold.major
Gemini 1.0 launched2023-12launchMultimodal from the ground up; claimed GPT-4 parity or superiority.major
Merger with Google Brain ends DeepMind's independent governance2023-04pivotDriven by ChatGPT competitive pressure, resource efficiency, product integration, and talent retention.major
Sparrow dialogue agent published2022publicationApplied RLHF and rule-based reward modeling with debate-style methods toward scalable oversight.moderate
AlphaFold protein database release2021launch200M+ protein structures freely available; accelerated global research.major
AlphaFold 2 reaches ~90% accuracy at CASP142020milestoneAddressed a 50-year grand challenge in protein folding.major
AlphaFold wins CASP132018milestoneFirst place in protein-prediction competition; proof of concept for AI in structural biology.major
AlphaZero generalizes to chess and shogi2017milestoneDomain-general strategic reasoning.major
AlphaGo Zero defeats AlphaGo 100-0 via self-play2017milestoneLearning without human data.major
AlphaGo defeats Lee Sedol 4-12016-03milestone200M+ viewers; demonstrated strategic AI; "Move 37" exemplified unexpected AI strategy.major
Acquired by Google2014acquisitionAcquisition structured to preserve DeepMind's autonomy — separate brand, ethics board for AGI oversight, open research publication, and UK headquarters.major
DeepMind founded in London2010foundingFounded by Demis Hassabis, Shane Legg, and Mustafa Suleyman with the mission to "solve intelligence, then use that to solve everything else."major

Divisions

1
Team

DeepMind's AI safety research team, focused on alignment, interpretability, and responsible development

Prediction Markets

21 active

Related Wiki Pages

Top Related Pages

Analysis

Anthropic Impact Assessment ModelAI Safety Multi-Actor Strategic Landscape

Policy

California SB 53Safe and Secure Innovation for Frontier Artificial Intelligence Models Act

Concepts

Scientific Research CapabilitiesAGI TimelineAgi DevelopmentLarge Language Models

Risks

AI Development Racing DynamicsReward Hacking

Other

InterpretabilityRLHFNeel NandaGeminiGemini 1.0 Ultra

Key Debates

Corporate Influence on AI PolicyAI Accident Risk CruxesWhy Alignment Might Be Hard

Historical

Deep Learning Revolution EraInternational AI Safety Summit Series