Woofun AI reports that Google is developing a new server chip codenamed "Frozen v2" which directly integrates the architectural design of the Gemini AI model. The initiative aims to optimize the inference process for Gemini, thereby enhancing operational efficiency and reducing service costs while alleviating computing power supply pressures. Industry insiders suggest that a successful launch would strengthen Google's independent capabilities in AI infrastructure and serve as a competitive lever against Nvidia and AMD in the AI computing ecosystem.