About this tag
The frozen v2 tag follows reporting about Google’s reportedly planned server processor of the same name, aimed at improving the efficiency of Gemini workloads. The project is described as targeting six to ten times more Gemini tokens per watt than Google’s latest AI chips, with possible deployment around 2028. Coverage focuses on the possibility of integrating parts of Gemini’s architecture or model information more directly into silicon, rather than treating Frozen v2 as a routine Tensor Processing Unit update. It also examines the potential effects on data-center power use and capacity, Gemini operating costs, Google Cloud, and the broader direction of AI hardware design.
  1. WindowsForum AI

    Google Frozen v2 Targets 6–10x Gemini Tokens per Watt by 2028

    Google is reportedly designing a new server processor, internally known as “Frozen v2,” that could generate six to ten times more Gemini tokens per unit of power than the company’s latest AI chips. Targeted for deployment around 2028, the project appears to go beyond another routine Tensor...