Google Develops New AI Chip to Boost Gemini Efficiency

Alphabet is developing a new in-house AI server chip designed to make Google’s Gemini artificial intelligence models significantly more efficient, according to a new report. The processor, internally known as Frozen v2, is expected to enter production in 2028 and could dramatically improve the company’s AI infrastructure while reducing computing costs.

According to The Information, the chip may deliver between six and 10 times the efficiency of Google’s current AI accelerators when measured by the number of AI tokens generated for each unit of power consumed. Improving energy efficiency has become a major priority as AI companies continue to expand large-scale model deployment and face growing demand for computing resources.

Google did not confirm specific details about the project but said its engineering teams are continually researching new hardware and software technologies. The company noted that not every experimental project reaches production. Still, its strategy of co-designing chips and software helps ensure its AI systems are highly optimised for real-world workloads.

Custom AI hardware race accelerates.

Major AI developers are increasingly investing in custom silicon to improve performance and reduce their dependence on external suppliers. The move also comes as the industry continues to deal with shortages of high-performance AI hardware and rising infrastructure costs.

For years, Nvidia has dominated the AI chip market, making many leading AI companies heavily reliant on its graphics processors. Developing proprietary chips allows companies such as Google to tailor hardware specifically for their own models while potentially lowering long-term operating expenses.

The trend has accelerated across the industry. OpenAI recently introduced its first custom inference processor, known as Jalapeño, while reports suggest Anthropic is exploring a chipmaking partnership with Samsung to strengthen its own AI infrastructure.

Efficiency becomes key to AI investment strategy.

Google’s investment in Frozen v2 also comes as investors closely monitor Alphabet’s substantial AI spending. Earlier this year, the company announced plans to invest between $180 billion and $190 billion as it expands its artificial intelligence capabilities across products and cloud services.

With such significant capital commitments, Alphabet faces increasing pressure to demonstrate that its AI investments will generate stronger performance while improving operational efficiency. Reports of the new chip appeared to reassure investors ahead of the company’s upcoming earnings announcement, with Alphabet’s shares rising around 3% following publication of the report.