The new AI chip, called "Frozen v2," would embed parts of Gemini's architecture directly into the silicon, according to the ...
The project is reportedly part of Google's broader effort to reduce AI infrastructure costs and lessen its reliance on Nvidia's GPUs.