Google DeepMind
Google's unified AI lab; maker of Gemini
Google DeepMind is the merged research org (DeepMind + Google Brain) building the Gemini family, the only frontier lab that is fully vertically integrated on its own TPU accelerators rather than Nvidia GPUs. It trains and serves on Google Cloud TPUs (Trillium/Ironwood generations) and ships Gemini across Search, Workspace, and Android. In 2025-2026 it pushed multimodal, long-context, and agentic Gemini variants and AlphaFold/AlphaProof-style science work.
Flagship model
Gemini
Compute
In-house TPU (no Nvidia dependency)
Parent
Alphabet / Google
Edge
Full-stack vertical integration
How it fits the stack
Google DeepMind with what it depends on (above) and what it feeds (below). The figure renders as a crawlable diagram and upgrades to an interactive 3D graph as it scrolls into view.
Google DeepMind in the AI stack. Google DeepMind with its immediate upstream dependencies (top) and downstream dependents (bottom) in the AI value chain. Hover a node in 3D, or read the full relationships below.
Graph data (text) — 11 entities, 11 relationships
- Google DeepMind —depends on→ Google Cloud (GCP)
- Google DeepMind —hosts→ Google Cloud (GCP)
- Google DeepMind —uses→ JAX
- Google DeepMind —uses→ JAX / XLA
- Google DeepMind —uses→ Model Context Protocol (MCP)
- Google DeepMind —supplies→ Reddit / content licensing deals
- Figure / Humanoid Robotics —uses→ Google DeepMind
- Google DeepMind —competes with→ Anthropic
- Google DeepMind —partners with→ Content & IP Licensing Deals
- Google DeepMind —competes with→ OpenAI
- Google DeepMind —competes with→ Perplexity AI
Depends on ↑ · 6
Context — capital, rivals, policy · · 12