Google plans new chip to run Gemini models more efficiently, the Information reports

Chronological Source Flow
Back

AI Fusion Summary

Google is developing a new AI chip called Frozen v2 to run Gemini models more efficiently. According to reports from The Information and Reuters, this hardware embeds Gemini model elements directly into the silicon. This design promises efficiency gains of 6-10x compared to existing TPUs. The company expects to deploy this new chip as early as 2028, marking a significant shift in how Google integrates its AI models into hardware infrastructure.
Community Comments
Loading updates...
0