Google plans custom chip to run Gemini models faster

Anabelle Colaco
21 Jul 2026, 16:44 GMT+10

Google is developing a new server chip designed to run its Gemini artificial intelligence models more efficiently by embedding parts of the model directly into the hardware
The Alphabet-owned company expects the chip, informally known as "Frozen v2," to help ease AI computing capacity constraints that have created internal tensions and led Google Cloud to turn away some potential customer deals
The report said Google aims to deploy the chip as early as 2028, although engineers are still finalizing its design, including how much of the AI model will be hardwired into the chip

MOUNTAIN VIEW, California: Google is developing a new server chip designed to run its Gemini artificial intelligence models more efficiently by embedding parts of the model directly into the hardware, according to a report by The Information citing people familiar with the matter.

The Alphabet-owned company expects the chip, informally known as "Frozen v2," to help ease AI computing capacity constraints that have created internal tensions and led Google Cloud to turn away some potential customer deals, the report said.

The report said Google aims to deploy the chip as early as 2028, although engineers are still finalizing its design, including how much of the AI model will be hardwired into the chip.

According to the report, the chip could be six to 10 times more efficient than Google's latest custom AI chips based on the number of AI tokens served per unit of power.

"Our teams are constantly researching and experimenting with new innovations... By co-designing our hardware and software from the ground up, we ensure our systems are integrated and highly optimized," a Google Cloud spokesperson said.

The report said the "Frozen" project is intended to create a new family of in-house chips that would complement, rather than replace, Google's existing tensor processing units (TPUs).

Bloomberg News reported last week that Google had delayed the launch of its latest Gemini AI model after it failed to meet internal performance goals. The company is continuing work to improve the model's capabilities, particularly in coding.

Shares of Alphabet rose 3.3% in early trading following the report.