Google is working on a new Frozen v2 AI processor to accelerate Gemini

This is reported by TechCrunch. The release of the chip is scheduled for 2028. According to preliminary calculations, Frozen v2 could be 6-10 times more efficient than Google’s existing AI chips, if measured by the number of tokens generated per unit of energy consumed.

In a statement to reporters, Google representatives did not directly confirm the rumors, but did not deny them either. «Our teams are constantly researching and experimenting with new innovations to ensure maximum performance and efficiency for our users and customers,» Google said. They also emphasized that jointly developing hardware and software from scratch allows for systems to be highly optimized for real-world workloads.

Tech companies are increasingly building their own chips, seeking to reduce their reliance on Nvidia and address the computing power shortage. Energy efficiency and cost control have become key factors in the industry as the excessive financial appetites of AI projects have somewhat dampened initial investor optimism.

For Alphabet, its own component base is an important part of the strategy that must justify large-scale investments in artificial intelligence infrastructure — at the beginning of the year, the company announced plans to spend between $180 billion and $190 billion.

Read the country’s main IT news in our Telegram

On the topic

Read the country’s main IT news in our Telegram

Google has integrated computer control directly into the new Gemini 3.5 Flash model

On the topic

Google has integrated computer control directly into the new Gemini 3.5 Flash model

Meta to start producing its own AI chip in September to reduce dependence on Nvidia and AMD

On the topic

Meta to start producing its own AI chip in September to reduce dependence on Nvidia and AMD

Chrome tests Gemini feature disable button after backlash over 4GB AI model

On the topic

Chrome tests Gemini feature disable button after backlash over 4GB AI model