Google Reportedly Developing 'Frozen v2' AI Server Chip for Gemini

 

Data center server racks used to illustrate AI computing infrastructure.
Photo Credit:TechCrunch 

Alphabet, Google's parent company, is designing a custom server processor internally codenamed "Frozen v2" to improve the efficiency of its Gemini AI models, with a reported target of 2028.

According to The Information, internal projections suggest the chip could deliver between six and 10 times greater efficiency than Google's existing custom AI chips, measured by tokens generated per unit of power.

In a statement to TechCrunch, Google did not directly confirm the reported project. A company spokesperson said that Google continues to research new hardware and software designs, adding that not every internal project reaches production.

Alphabet previously projected capital expenditures between $180 billion and $190 billion this year. Alphabet shares were up about 3% after the report.

This report is based on reporting by TechCrunch and The Information.

  • Primary Source: TechCrunch — Google is working on a new AI chip designed to make Gemini more efficient (Published July 20, 2026).

  • Secondary Source: The Information (Internal project reporting and efficiency metrics).

  • Post a Comment

    0 Comments