Google DeepMind

What Google DeepMind does

Google DeepMind is the AI research and product division of Alphabet Inc., headquartered in London (with a new 330-metre King's Cross HQ) and Mountain View, and operated through regional hubs including Singapore. Originally founded in 2010 as DeepMind, acquired by Google in 2014 for approximately $500 million, and merged with Google Brain in April 2023, the organization builds next-generation AI systems spanning the Gemini 3.5 family of natively multimodal foundation models (Flash, Pro, Deep Think, Flash-Lite), generative media models (Veo 3.1, Imagen, Lyria 3, Gemini Omni, Gemini Image/Nano Banana, Gemini Audio), world and embodied AI (Genie 3, Gemini Robotics, SIMA 2), open models (Gemma 4), and scientific AI systems (AlphaFold, AlphaGenome, AlphaEarth, WeatherNext, AlphaEvolve, Co-Scientist). These capabilities were validated by the 2024 Nobel Prize in Chemistry awarded to Demis Hassabis and John Jumper for AlphaFold, which has predicted more than 200 million protein structures.

The organization serves four primary customer sets: global financial institutions led by HSBC's multi-year partnership covering 200+ AI use cases; software developers and developer-tools platforms including GitHub, Cursor, Replit, JetBrains, and Figma; large enterprise SaaS and data platforms such as Salesforce, Shopify, Box, Databricks, Xero, and Thomson Reuters; and mass-market consumers via the Gemini app, Google Search AI Mode, Workspace, YouTube, and creative tools like Google Flow. Distribution is API-first through ai.google.dev and Google AI Studio, complemented by the Gemini Enterprise Agent Platform on Google Cloud, OEM/embedded integrations across Alphabet products and third-party SaaS, and consumer subscription.

Monetization spans subscription (Gemini app Google AI subscription), usage-based API billing, multi-year enterprise platform contracts (e.g., HSBC), and strategic equity investments such as the $75 million Alphabet stake in A24. As a wholly owned Alphabet division, DeepMind is funded by its parent and does not disclose standalone revenue. Recent notable activities include the publication of a 35-page AI Control Roadmap for AI agent security, a UK government partnership on AI-accelerated planning tools, the launch of Gemini 3.5 and Veo 3.1, and the opening of new offices in Singapore and London, set against high-profile talent departures to Anthropic and OpenAI in mid-2026.

Name

Google DeepMind

Legal name

Google DeepMind

Website

https://deepmind.google

Company type

Private

Founded year

2010

Operating status

Operating

Headcount range

5,001–10,000 employees

Short description

Google DeepMind is Alphabet's AI research division building frontier multimodal foundation models (Gemini 3.5), generative media (Veo, Imagen, Lyria), and scientific AI (AlphaFold) for global enterprises, developers, consumers, and scientists.

Ownership category

akta.pro rank

Product category

Artificial Intelligence Foundation Models

NAICS

Software Publishers (51321), Computer Systems Design and Related Services (54151)

SIC

Services-Computer Integrated Systems Design (7373), Services-Computer Programming Services (7371)

akta.pro primary industry

End-to-End Enterprise AI Platforms (MLOps & Model Lifecycle Management) (HDAEANAA)

akta.pro secondary industries

Enterprise Foundation Model Integration & APIs (Connectors, Governance, Deployment) (HDAAACAO), Responsible AI, Security & Privacy Platforms (Safety, Guardrails, PII) (HDAEANAG), AI Integration & Orchestration Platforms (Connectors, Workflow, iPaaS for AI) (HDAEANAI)

Keywords

• Foundation AI models
• Generative AI platforms
• Multimodal AI systems
• Enterprise AI agents
• AI research services

Headquarters

HQ city

London

HQ country

United Kingdom

HQ region

Europe

Offices4 records

Markets served

GTM type

B2B and B2C

Offering type

Software

Cost components

Technology or R&D, Personnel, Infrastructure, Marketing or Sales, Operations

Revenue model

1. Consumer subscription (Gemini app): Google AI subscription required for premium Gemini app features; tiered pricing with features varying by tier and geography.
2. API / usage-based developer access: Revenue from Gemini API (ai.google.dev) billed on usage (tokens/compute) for developers integrating Gemini, Imagen, Veo, Lyria, Gemma, and other models into applications.
3. Enterprise AI platform contracts: Multi-year enterprise agreements via Google Cloud and the Gemini Enterprise Agent Platform, exemplified by the HSBC partnership targeting over 200 AI use cases with each initiative estimated to return more than US$100 million in revenue or efficiency gains for the customer.
4. Strategic equity investments: Strategic equity stakes in aligned companies, such as Alphabet/Google's $75 million investment in A24 (approximately a 2% stake at a $3.5 billion valuation).

Pricing tiers

Go-to-market motion5 records

Distribution channels9 records

Marketing channels11 records

Core offering

Google DeepMind researches and builds frontier AI systems — primarily the Gemini family of multimodal foundation models — and delivers them to consumers via the Gemini app, to developers via Google AI Studio, the Gemini API and Google Antigravity, and to enterprises via the Gemini Enterprise Agent Platform on Google Cloud. The portfolio extends into scientific AI (AlphaFold, AlphaGenome, AlphaEarth, WeatherNext, AlphaEvolve, Co-Scientist), generative media (Veo, Imagen, Lyria, Gemini Audio, Gemini Omni, Nano Banana), world and embodied AI (Genie 3, Gemini Robotics, SIMA 2), open models (Gemma), and provenance tooling (SynthID).

Product overview

Google DeepMind operates a unified AI research and product portfolio built around its flagship Gemini family of foundation models, which serves as the central platform layered with specialized models and tools. The Gemini 3.5 family (Pro, Flash, Flash-Lite, Deep Think) powers core text, reasoning, and agentic capabilities, while the broader portfolio spans Gemini Omni (multimodal video generation), Gemini Image / Nano Banana (image generation), Gemini Audio (live dialogue, TTS, and real-time translation), Veo (cinematic video with audio), Imagen (text-to-image), Lyria (music), Gemma (open models), SynthID (watermarking), and Gemini Robotics / Genie 3 (world models and embodied AI). On the science side, AlphaFold, AlphaEarth, WeatherNext, AlphaEvolve, and AlphaGenome apply foundation-model techniques to scientific discovery. These models are consumed through consumer products (the Gemini app), developer surfaces (Google AI Studio, Google Antigravity, the Gemini API / Gemini Live API), and the Gemini Enterprise Agent Platform for enterprise agent deployment. The company's mission is to build AI responsibly to benefit humanity, with an accompanying Responsibility & Safety program that includes the AI Control Roadmap, multi-agent safety research, and SynthID-based content provenance.

Differentiator

Problem solved

Functional benefit

Brands

• Gemini: Frontier AI model family and conversational assistant
• AlphaFold
• Veo
• Imagen
• Lyria
• Gemma
• SynthID
• Genie
• Gemini Robotics
• WeatherNext
• AlphaEvolve
• AlphaEarth

Products and services

• Gemini Google DeepMind's flagship family of frontier foundation models (Gemini 3.5 series, Gemini 3.1 Pro, Gemini 3.1 Flash-Lite, Gemini 3.1 Deep Think) that combine text, image, video, and audio understanding with agentic action capabilities for coding, reasoning, and long-horizon workflows.
• Gemini Omni Multimodal generative model that creates video from any combination of inputs (text, image, video, audio), with world understanding, multi-turn editing, motion/style transfer, and real-world physics grounding. Outputs are watermarked with SynthID and C2PA Content Credentials.
• Nano Banana (Gemini Image) Family of native multimodal image generation and editing models (Nano Banana 2 built on Gemini 3.1 Flash Image; Nano Banana Pro built on Gemini 3 Pro Image) that capture prompt nuance through deep language understanding and support conversational, multimodal image creation.
• Gemini Audio Audio model family including 3.1 Flash Live (low-latency live dialogue), 3.5 Live Translate (real-time speech-to-speech translation across 70+ languages), and 3.1 Flash TTS (expressive speech generation with intonation/inflection control), all with SynthID watermarking.
• Veo Leading video generation model family (Veo 3.1) producing cinematic video with native audio (sound effects, ambient noise, dialogue), 1080p/4K resolution, and creative controls including reference images, style transfer, character consistency, scene extension, and camera/object controls.
• Imagen Google DeepMind's leading text-to-image model for generating high-quality images from text prompts.
• Lyria Music generation model (Lyria 3) that composes music with vocals and supports experimentation with acoustic details and real-time collaboration.
• Genie 3 World model that generates and explores interactive 3D worlds, representing a new frontier for world models that simulate environments.
• Gemini Robotics Most advanced vision-language-action model for embodied AI, enabling robots to perceive, reason, use tools, and interact with the physical world.
• Gemma Open model family (Gemma 4, including the 12B encoder-free multimodal variant) built to maximize intelligence-per-parameter, designed for responsible AI applications at scale and available via Google AI Studio.
• Gemini Embedding State-of-the-art multimodal embedding model (Gemini Embedding 2) for semantic search and RAG workflows, available via the Gemini API and Google Cloud Vertex AI.
• SynthID Watermarking technology that embeds imperceptible digital signatures in content (images, audio, video, text) generated by Google's AI models to identify AI-generated content; outputs from Omni, Veo, Gemini Audio, and Gemini Image are automatically marked with SynthID and C2PA Content Credentials.
• AlphaFold AI system that predicts protein 3D structures with high accuracy from amino acid sequences; AlphaFold 2 solved the long-standing protein structure prediction problem and has predicted over 200 million structures available via the AlphaFold Protein Structure Database.
• WeatherNext Fast and accurate AI weather forecasting model for predicting atmospheric conditions.
• AlphaEarth Satellite-data foundation model that maps the planet in unprecedented detail using geospatial AI.
• AlphaEvolve Gemini-powered coding agent that designs advanced algorithms for math and computing applications.
• AlphaGenome AI model for better understanding the human genome.
• SIMA 2 An agent that plays, reasons, and learns with humans in virtual 3D worlds, representing a generalist embodied agent.
• Co-Scientist A multi-agent AI research partner designed to accelerate scientific research by collaborating with researchers.
• Gemini app Consumer chat application providing access to Gemini models for productivity and creativity use cases.
• Google AI Studio Developer environment to build with next-generation Gemini models and tools, providing the fastest path from prompt to production.
• Google Antigravity AI-first agentic development platform that allows anyone to be a builder; Antigravity 2.0 released in May 2026.
• Gemini Enterprise Agent Platform Enterprise platform on Google Cloud for building, scaling, and governing AI agents across business workflows.

Quantifiable outcome

• 42% better than Flash 3 on Armadin's long-range multi-turn cyber benchmark, with 68% improvement in token efficiency
• +6 more outcomes

Named customers22 records

Segments7 records

Ideal customer profiles7 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration1 record

AI capability17 records

Feature17 records

Partnerships

Eight partnerships are on record, tiered flagship, core and minor.

• A24flagshipStrategic or Co-development Partner · 22 June 2026Multi-year, nonexclusive AI research partnership announced June 22, 2026, backed by an approximately $75 million equity investment from Alphabet/Google (roughly a 2% stake at A24's $3.5 billion 2024 valuation). The arrangement develops AI tools for film production — including AI-generated storyboards and production planning — with researchers working alongside A24 artists. A24 retains full creative control; the deal explicitly excludes any transfer of A24's film/TV library or use of content for AI training.
• HSBCflagshipStrategic or Co-development Partner · 17 June 2026Multi-year strategic partnership announced at Google Cloud Summit London 2026 with Google Cloud and Google DeepMind to deploy AI capabilities across HSBC's global operations, targeting over 200 new AI use cases over two years. Each priority initiative is estimated to return more than US$100 million in revenue gains or efficiency improvements. Focus areas: hyper-personalized wealth management, financial crime risk management (monitoring nearly 1 billion monthly transactions), and AI-empowered frontline staff.
• UK Ministry of Housing, Communities and Local Government (MHCLG)coreStrategic or Co-development Partner · 17 June 2026Collaboration with Google Cloud, Google DeepMind, and Faculty to develop the Augmented Planning Decisions (APD) AI tool, currently being alpha-tested with Barnet Council, Dorset Council, and Camden Council, with plans for national deployment from 2027. MHCLG and i.AI incubator are also rolling out Extract AI to all councils in England using DeepMind's Gemini and Meta's Segment Anything Model.
• Darren Aronofsky's Primordial SoupcoreStrategic or Co-development PartnerPartnership shaping Veo's capabilities to open new possibilities for cinematic storytelling. Primordial Soup is using Veo to explore new filmmaking techniques — including how to integrate live-action footage with Veo-generated video — and has produced three short films with emerging filmmakers through the partnership.
• Google CloudcoreTechnology or IntegrationGoogle Cloud serves as the primary enterprise distribution and infrastructure partner for DeepMind models, including the Gemini Enterprise Agent Platform. DeepMind engineering teams work alongside Google Cloud forward-deployed engineers on enterprise deals like HSBC. DeepMind models are accessible via Google Cloud's model garden and Gemini Enterprise Agent Platform.
• Accenture (Faculty)minorImplementation/ SI/ Consulting PartnerFaculty (acquired by Accenture) is collaborating with Google Cloud, Google DeepMind, and MHCLG on the Augmented Planning Decisions (APD) AI tool for UK council planning applications.
• 15 European robotics startupscoreOthersGoogle DeepMind launched an AI robotics accelerators program supporting 15 European robotics startups through mentorship, technical resources, and access to advanced AI tools. The initiative spans manufacturing, healthcare, logistics, construction, and industrial automation, bridging AI research and real-world deployment.
• U.S. Government (voluntary AI safety framework)coreOthersGoogle DeepMind is among major AI companies (with OpenAI, Anthropic, Microsoft, xAI) that have committed to providing early access to frontier models for U.S. national security evaluations under the Trump administration's voluntary framework established by a June 2 executive order, before wider deployment.

Scale indicators11 records

Recent moves6 records

Expansion highlights6 records

Emerging players

• Mistral AI: European frontier/open-weight lab offering competitive open and commercial models with strong enterprise traction. Competes with Gemma and Gemini Flash in cost-efficient inference and with European sovereignty buyers.
• Cohere: Enterprise-focused foundation model provider offering Command, Embed, and RAG products. Competes with Gemini Embedding and Gemini API for enterprise NLP and retrieval workloads, especially in regulated industries.
• Perplexity AI: AI-native answer engine built on top of multiple frontier models. Competes with Gemini-in-Google-Search AI Mode for consumer search and information-retrieval use cases.

Direct peers

• Anthropic: Direct competitor in frontier models (Claude family) and AI agent safety. Poached DeepMind VP John Jumper and researchers Adler and Pritzel, and competes head-to-head for enterprise agent deals on AWS Bedrock and Google Cloud.
• xAI: Elon Musk's frontier AI lab (Grok models) and a participant in the same U.S. voluntary AI safety framework as DeepMind. Competes for top research talent and direct-to-consumer AI share via the X platform.
• OpenAI: Direct competitor in the frontier foundation-model race, offering GPT/ChatGPT, the OpenAI API, and the Operator agent platform. Both labs target enterprise and developer customers with comparable multimodal, agentic, and reasoning capabilities.
• Meta AI (FAIR): Develops the Llama open-weight family and the Segment Anything Model (jointly used with DeepMind in the UK Extract AI tool). Competes with Gemma as an open-model alternative and with closed frontier models via Meta AI assistants.

Broad incumbents

• Microsoft AI: Distributes OpenAI models via Azure and Copilot, and is a participant in the U.S. voluntary AI safety framework. Competes with the Gemini Enterprise Agent Platform on Google Cloud for enterprise agent workloads.
• Amazon (AWS AI / Bedrock / Anthropic): Hosts Anthropic Claude on Bedrock and offers its own Titan models, competing with Gemini on Google Cloud for enterprise AI infrastructure and managed model services.

Regional players

• Baidu (ERNIE / Baidu AI Cloud): China-based frontier AI lab with ERNIE foundation models and AI Cloud. Competes regionally with DeepMind's Asia-neutral Singapore hub in the Chinese enterprise and government markets.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat8 records

Key risks6 records

Key highlights7 records

Customer concentration

• Xx.com/deepmindai
• GitHubgithub.com/google-deepmind
• YouTubeyoutube.com/@GoogleDeepMind
• LinkedInlinkedin.com/company/googledeepmind
• Instagraminstagram.com/googledeepmind

Revenue estimate

Valuation estimate

Number of profiles

Profiles5 records

Subsidiaries1 record

Funding overview

Funding rounds1 record

Investors2 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A3 records

Investments9 records

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Google DeepMind