BREAKING
Google's 'Frozen' Chip to Cut Gemini Cost
0x
low estimate
0x
high estimate
0
possible deployment
How 'Frozen' Works
1Freeze architecture in silicon
2Reduce data movement
3Keep weights updatable
4Speed up inference
Frozen v2 vs Existing TPUs
Frozen v2new line
Model-specific ASIC
Experimental scale
Targets Gemini inference
TPUssince 2016
General-purpose accelerator
8th-gen lineup April 2026
Training and inference
Split Reaction, Unconfirmed Plan
Projections, Not Measured Results
AI NEWS BLITZ
A report says Google is building a chip that hardwires Gemini into silicon to slash inference costs.