Google Launches Frozen v2 AI Chip with 6-10x Power Efficiency for Gemini

Google's new Frozen v2 chip directly embeds Gemini architecture into silicon, aiming to reduce calculations and data movement for AI queries. This specialized chip is projected to deliver 6-10 times more tokens per unit of power than current TPUs, with deployment targeted for 2028 to alleviate Google's compute capacity constraints.

View Tweet→

Sources (1)

Semicon Twitter · Twitter · 1d ago

View tweet→

Credibility

Lead sourceSemicon Twitter

Corroboration1 distinct outlet

VerificationSingle source

Entities

Alphabetsubject

Unlock the full signal with Pro

Tessara synthesizes 80+ data sources into actionable intelligence for AI infrastructure investors.

Unlock with Pro→Sign in

← SignalsSignal

Watch7.5LaunchComputeSingle sourceactive

Updated 1d ago · 1 source

Google Launches Frozen v2 AI Chip with 6-10x Power Efficiency for Gemini

Google's new Frozen v2 chip directly embeds Gemini architecture into silicon, aiming to reduce calculations and data movement for AI queries. This specialized chip is projected to deliver 6-10 times more tokens per unit of power than current TPUs, with deployment targeted for 2028 to alleviate Google's compute capacity constraints.

View Tweet→