Google is reportedly building a Gemini-specific chip for far better efficiency
mark_k · x · 2026-07-20
Google is reportedly working on a new AI server chip, informally called Frozen v2, that would embed parts of Gemini directly into hardware.
Reported details
- The chip is said to target 6–10× more tokens per watt than Google’s latest custom AI chips.
- It is intended to complement TPUs, not replace them.
- Deployment is reportedly aimed at 2028.
The broader implication is that AI chips may increasingly be co-designed with the models they run, blurring the line between software and silicon.
Related event: Google Reportedly Developing Gemini-Specific Chip Frozen v2(10 posts)→
More from Infra
- Grid queues make power architecture an AI-factory decision; Exowatt and Redwood offer fixes — ingliguori · 2026-10-27
- Two CVSS 9.9 Wasmtime flaws allow sandbox escape, one ARM64-only since 2020 — basedjensen · 2026-10-27
- AI Infra Meme: While Countries Need TW-Scale Power for Datacenters, People Argue About Lightbulbs — mgostIH · 2026-10-27
- NVIDIA reportedly stops taking RTX 5090 orders from board partners, even backorders — chemist_slime · 2026-10-27
- Legora runs 2,000 parallel agent evals on Modal Sandboxes for legal AI — blelbach · 2026-10-27
- Cloud release velocity scores: AWS 95, Azure 90, Google Cloud 82 — DavidLinthicum · 2026-10-27