According to two people familiar with the matter, Google is developing a new server chip that can directly solidify the underlying architecture of the Gemini model into the chip hardware, thereby greatly improving the operational efficiency of providing AI services to users. This internal chip, codenamed Frozen v2, aims to solve Google's current serious shortage of AI computing power. The computing power gap has caused internal resource conflicts, and even forced Google Cloud to reject cooperation orders from many external customers. According to people familiar with the matter, the R&D team estimates that after the chip is officially launched, the number of words that can be processed per unit of power will reach 6 to 10 times that of Google's current self-developed AI chip products, achieving a qualitative leap in energy efficiency. The origin of the code name “Frozen” stems from its core design idea: to permanently etch part of the computational logic of the big model into silicon hardware. Engineers are still finalizing the core functional modules of this chip, as well as a scheme for the components to work together. People familiar with the matter said that Google plans to deploy the chip as early as 2028.

According to two people familiar with the matter, Google is developing a new server chip that can directly solidify the underlying architecture of the Gemini model into the chip hardware, thereby greatly improving the operational efficiency of providing AI services to users. This internal chip, codenamed Frozen v2, aims to solve Google's current serious shortage of AI computing power. The computing power gap has caused internal resource conflicts, and even forced Google Cloud to reject cooperation orders from many external customers. According to people familiar with the matter, the R&D team estimates that after the chip is officially launched, the number of words that can be processed per unit of power will reach 6 to 10 times that of Google's current self-developed AI chip products, achieving a qualitative leap in energy efficiency. The origin of the code name “Frozen” stems from its core design idea: to permanently etch part of the computational logic of the big model into silicon hardware. Engineers are still finalizing the core functional modules of this chip, as well as a scheme for the components to work together. People familiar with the matter said that Google plans to deploy the chip as early as 2028.