Alphabet Inc. (NASDAQ:GOOGL) shares rose 1.15% on Monday after a report said Google is developing a new generation of AI chips that could dramatically improve efficiency by embedding its Gemini artificial intelligence model directly into the chip architecture.
According to The Information, the experimental initiative, known internally as “Frozen v2,” is intended to eliminate much of the processing overhead associated with conventional GPUs and TPUs. Rather than relying on general-purpose hardware, the design incorporates Gemini’s AI architecture directly into the silicon, reducing both computational steps and data movement.
Google targets major efficiency gains
The report said the proposed chip could deliver between six and ten times greater power efficiency per token than Google’s current TPU systems.
The project is also intended to address growing demand for AI computing resources, helping ease internal capacity constraints while allowing Google Cloud to support more enterprise customers.
Current plans reportedly target deployment as early as 2028, with the technology expected to complement rather than replace Google’s existing TPU product family.
While the hardware would be optimised around Google’s current AI architecture, engineers would still be able to update the system through new model weights without redesigning the silicon itself.
Google says research projects may not reach production
Responding to the report, a Google spokesperson said the company regularly explores new approaches to improve AI efficiency but cautioned that not every experimental project ultimately becomes a commercial product.
The reported initiative highlights Google’s continued investment in custom AI hardware as major technology companies race to develop more efficient infrastructure capable of supporting increasingly complex artificial intelligence workloads.
Alphabet stock price
