Google is reportedly working on a next-generation custom artificial intelligence (AI) server chip designed specifically to power its Gemini family of AI models, marking another major step in the global race for AI infrastructure. The new chip, internally codenamed “Frozen v2,” is expected to significantly improve how efficiently Gemini processes AI workloads while reducing Google’s dependence on third-party graphics processors, particularly those supplied by Nvidia. The development reflects Google’s long-term strategy to strengthen its AI ecosystem through custom-built hardware and address growing demand for AI computing resources.
A Custom Chip Built for Gemini
According to reports, Google is designing the new server chip to integrate portions of the Gemini AI model directly into the hardware architecture.
Unlike general-purpose AI accelerators, this specialized chip would be optimized for AI inference—the process of generating responses after a model has already been trained. By tailoring the hardware specifically for Gemini, Google aims to deliver faster performance while consuming significantly less power.
Why Google Is Building Its Own AI Hardware
Over the past two years, demand for AI computing has surged as companies deploy increasingly capable large language models.
Most AI companies currently rely heavily on Nvidia’s GPUs to train and run these models. However, growing global demand has resulted in limited supply and soaring costs.
Developing proprietary AI chips would allow Google to:
- Reduce dependence on external suppliers.
- Lower long-term infrastructure costs.
- Improve energy efficiency.
- Optimize Gemini’s performance.
- Expand AI services more rapidly.
Industry experts view custom silicon as one of the biggest competitive advantages in modern artificial intelligence.
AI Capacity Crunch Inside Google
The report also highlights that Google’s AI infrastructure is facing unprecedented demand.
As Gemini adoption continues to grow across Google Search, Workspace, Cloud, Android, and developer services, computing resources have become increasingly constrained. Reports indicate that Google Cloud has even had to limit some external AI workloads because available computing capacity is being prioritized for Gemini and internal services.
Up to Ten Times More Efficient
One of the most notable claims is the chip’s expected efficiency.
The report suggests the new processor could become six to ten times more efficient than Google’s latest generation of AI chips when measured by the number of AI tokens generated per unit of power consumed.
If achieved, this would represent a major breakthrough in reducing the enormous energy costs associated with running modern AI systems.
Challenging the AI Hardware Leaders
Google’s move places it in direct competition with other technology giants investing heavily in custom AI hardware.
Companies including Microsoft, Amazon, Meta, Apple, and OpenAI are all expanding AI infrastructure, while Nvidia remains the dominant supplier of AI GPUs worldwide.
By developing specialized hardware, Google hopes to gain greater control over both AI performance and operating costs, strengthening Gemini’s competitiveness in the rapidly evolving AI market.
Market Reacts Positively
Investors responded positively to the reports.
Alphabet shares climbed following news of the project, reflecting investor confidence that custom AI chips could improve Google’s long-term profitability while strengthening its leadership in generative AI.
Looking Ahead
Although the chip is still under development and reports suggest deployment could occur later this decade, the project demonstrates Google’s commitment to building a vertically integrated AI ecosystem where both software and hardware are designed together.
As AI demand continues to grow worldwide, custom processors may become one of the defining competitive advantages among leading technology companies.
Conclusion
Google’s reported development of the Frozen v2 AI chip highlights the next phase of competition in artificial intelligence. Rather than relying solely on more powerful software models, major technology companies are increasingly focusing on designing specialized hardware that can deliver faster performance, lower costs, and greater energy efficiency. If successful, Google’s new chip could significantly strengthen the Gemini ecosystem while reshaping the future of AI infrastructure.