Alphabet, Google’s parent company, is reportedly developing a new server chip designed to improve the efficiency of its in-house Gemini AI models. The chip, internally referred to as “Frozen v2,” is expected to launch in 2028 and could be six to 10 times more efficient than Google’s existing AI chips based on tokens generated per unit of power, according to a report citing anonymous sources.
Google did not directly confirm or deny the report but said its teams continually research new hardware and software innovations aimed at improving performance and efficiency, emphasizing its approach of co-designing hardware and software together.
The move reflects a broader industry trend of AI companies developing custom chips to improve efficiency and reduce dependence on Nvidia, which has long dominated the AI chip market. Earlier this year, OpenAI unveiled its first custom chip, Jalapeño, while Anthropic was reported to be in discussions with Samsung on a custom chip partnership.
The chip news comes as Alphabet faces investor scrutiny over its planned AI infrastructure spending, projected between $180 billion and $190 billion. Following the report, Alphabet shares rose approximately 3%, suggesting investors viewed the efficiency gains positively ahead of the company’s upcoming earnings report.
