What's happened
Google is developing a proprietary chip, codenamed Frozen v2, aimed at far greater efficiency for Gemini models. The project, expected to deploy in 2028, could achieve six to ten times more tokens per unit of power compared with current TPUs, according to The Information. The move underscores Google’s push to own more of its AI stack amid supply constraints and the broader shift away from Nvidia hardware.
What's behind the headline?
Context and implications
- Google is pursuing deeper co-design of hardware and software to boost efficiency for Gemini models.
- Frozen v2 is positioned as a specialized accelerator rather than a drop-in replacement for TPUs.
- The move reflects broader industry pressure to secure AI compute amid supply constraints and Nvidia’s dominance.
Questions for readers
- Will Frozen v2 alter Google Cloud’s ability to meet enterprise demand?
- How might this affect Nvidia’s market position if Google accelerates in-house chip design?
Forecast
- Deployment in 2028 is aimed at easing compute shortages and improving energy efficiency, potentially reshaping enterprise AI procurement.
How we got here
The Information reports Google is pursuing a highly energy-efficient freeze-dried hardware approach to support its Gemini AI architecture, aiming to ease an internal compute shortage and reduce dependence on external chipmakers. The project is described as a test-bed rather than a full-scale replacement for TPUs.
Our analysis
- TechCrunch: reports on Frozen v2 and quotes from Google; - CNBC: details on architecture integration and deployment timeline; - The Information: primary source on the chip's efficiency goals and internal use.
Go deeper
- What impact will Frozen v2 have on Google’s cloud compute offerings?
- Will Nvidia respond to Google’s in-house chip strategy?
- When will more details from Google surface about scale and production?”],
- scoresdrillingsideways
More on these topics
-
Nvidia - Computer game company
Nvidia Corporation is an American multinational technology company incorporated in Delaware and based in Santa Clara, California.
-
OpenAI - Artificial intelligence company
OpenAI is an artificial intelligence research laboratory consisting of the for-profit corporation OpenAI LP and its parent company, the non-profit OpenAI Inc.
-
Google - Technology company
Google LLC is an American multinational technology company that specializes in Internet-related services and products, which include online advertising technologies, a search engine, cloud computing, software, and hardware.