Google is developing a new server chip, internally codenamed Frozen V2, designed to integrate the architecture of its Gemini AI model for significantly improved operational efficiency, according to reports from The Information.
The development underscores the tech giant's strategy to tailor hardware specifically for its generative AI workloads.
By aligning the chip's architecture directly with the Gemini model, Google aims to enhance performance and reduce the computational costs associated with running large language models at scale.
This move comes as major technology companies increasingly seek to control their semiconductor supply chains to mitigate bottlenecks and optimize AI inference.
While Nvidia remains the dominant supplier of AI accelerators, firms like Google, Amazon, and Microsoft have invested heavily in custom silicon to gain a competitive edge in efficiency and cost.
The report adds to a broader narrative of consolidation and innovation in the AI hardware sector.