344
Productivity & Workflow355
Automation & Workflow224
Software Development251
Marketing & Growth192
AI Infrastructure & MLOps174
Writing & Content Creation203
Data & Analytics141
Design & Creative170
Photography & Imaging156
Customer Support131
Sales & Outreach125
Voice & Speech135
Education & Learning131
Operations & Admin87
A report says Google is working on a server chip called Frozen v2 to make its Gemini AI use less power and money. It could arrive in 2028.
In short: A report says Google is building a new server chip to help its Gemini AI run using less electricity.
Alphabet, Google’s parent company, is reportedly designing a new chip for servers (the powerful computers that run online services). The chip is meant to make Google’s Gemini AI models work more efficiently, meaning it can produce more output while using less power.
The chip is reportedly called “Frozen v2” inside Google and could be released in 2028, according to The Information, which cited anonymous sources. The report says it could be about six to 10 times more efficient than Google’s current AI chips, based on how many “tokens” it can generate per unit of power. Tokens are small pieces of text that AI works with, like words or parts of words (think of them like the beads an abacus counts).
Google did not directly confirm the report when asked by TechCrunch. It said its teams are always researching new ways to improve performance and efficiency, and that not every project becomes a real product.
Running AI costs a lot of money and electricity, especially for large services like Gemini. If Google can do the same work with less power, it can lower its operating costs and reduce pressure on limited chip supply. It can also rely less on Nvidia, the company that has dominated the market for the chips many AI systems use.
The report also appeared to calm some investor worries about Google’s large AI spending plans. After the report was published, Alphabet’s stock rose by about 3% on Monday morning ahead of its earnings report.
Source: TechCrunch AI