Skip to content
AI IntelligenceJul 20, 2026AI Intelligence
Article

Google is developing a new server chip called 'Frozen v2' that bakes the Gemini architecture directly into hardware.

It is said to be 6 to 10 times more efficient than current TPUs and could drastically reduce AI inference costs from 2028.

Data Cube AI EditorialSource: The Decoder
01

Source Brief

Google is developing a new server chip called 'Frozen v2' that bakes the Gemini architecture directly into hardware. It is said to be 6 to 10 times more efficient than current TPUs and could drastically reduce AI inference costs from 2028.

02