Google’s Unreleased Chip Could Run Gemini 6–10 Times More Efficiently, Report Says
Google is developing a server chip called 'Frozen v2' that could run Gemini 6–10 times more efficiently than its current AI chips, according to The Information. Slated for 2028 release, the move aligns with broader industry efforts to reduce dependence on Nvidia.
Alphabet is designing a server chip aimed at making its Gemini AI models run more efficiently, according to reporting by The Information citing anonymous sources. The chip, internally called "Frozen v2," is slated for release sometime in 2028 and could deliver between six and 10 times greater efficiency than Google's current AI chips when measured by tokens generated per unit of power.
Google did not directly confirm the report when reached by TechCrunch, though it declined to deny it. The company said in a statement: "Our teams are constantly researching and experimenting with new innovations to deliver maximum performance and efficiency for our users and customers. While not every project moves into production, this rigorous exploration is central to our full stack approach. By co-designing our hardware and software from the ground up, we ensure our systems are integrated and highly optimized for real-world workloads."
The development reflects a broader industry trend. AI companies including OpenAI and Anthropic have begun building custom chips to run their models more efficiently and address global shortages in AI computing capacity. OpenAI launched its first custom inference processor, called Jalapeño, in June. Anthropic is reportedly in talks with Samsung on a chipmaking partnership.
This move carries strategic weight for Google. Alphabet announced plans earlier this year to spend between $180 billion and $190 billion on its AI infrastructure and strategy. With such substantial investment on the line, the company needs to demonstrate concrete efficiency gains. The news of Frozen v2 appeared to reassure investors—Alphabet's stock climbed roughly 3% on Monday morning following The Information's publication of the report.
The push for custom chips also reflects an effort across the industry to reduce reliance on Nvidia, which has historically dominated the AI chip market. Concerns about AI spending have also shifted focus away from raw capability toward operational efficiency as a key selling point.
