Listen to the article: The audio text is automatically generated by an automated system

0:00

1 minute to read

Google is working on developing a server chip that would integrate elements of the Gemini model directly into the chip, in an attempt to deliver its artificial intelligence models to users more efficiently, according to what the news site "The Information" reported on Monday, citing people familiar with the matter.

The subsidiary of Alphabet expects that the new chip, named "Frozen v2," will help address the crisis of shortage of computing power for artificial intelligence, a crisis that has fueled internal tensions and pushed Google Cloud to reject deals with external customers.

"Google" plans to deploy the new chip as early as 2028, although engineers are still finalizing its design and determining how much model-specific information will be permanently integrated into the chip, according to what "Reuters" reported from "The Information."

The chip could be six to ten times more efficient than Google's latest dedicated AI chips, based on the number of AI tokens that can be processed per unit of energy consumed.

The "Frozen" project aims to develop a new family of internally designed chips, alongside Google's Tensor Processing Units, not to replace them.

Bloomberg reported last week that Google delayed the launch of its latest AI model "Gemini" after it failed to meet the company's internal targets, and it is currently working on improving its capabilities, especially in programming.

Advertorial

Advertorial

Read also

"Google" launches "search agents" in AI mode to track information on your behalf; currently available to all AI Ultra subscribers; applications and programs

Serious vulnerability in Gemini allows sending messages without knowing the lock code; threatens phones running Android 16; applications and programs

Major publishers accuse Gemini of training on copyrighted books; "Google" used copies of books from Google Books; Tech angles