Google's parent company, Alphabet, is reportedly in the advanced stages of developing a new in-house artificial intelligence (AI) chip, codenamed "Frozen V2." This strategic move is intended to significantly enhance the performance and efficiency of Google's Gemini AI models, with manufacturing projected to commence in 2028.
According to reports, the Frozen V2 chip is designed to be substantially more efficient than Google's current AI hardware. It is expected to function as a dedicated AI accelerator, capable of generating 6 to 10 times more AI tokens for the same amount of electricity. This breakthrough could translate into faster AI performance, reduced power consumption, and lower operational costs for Google's expansive AI infrastructure.
Boosting Gemini and Gaining a Competitive Edge
The development of Frozen V2 comes at a critical time for Google. The company has reportedly faced performance challenges with its flagship AI model, Gemini 3.5 Pro, leading to delayed public releases. The new chip is envisioned to resolve such bottlenecks, providing a dedicated hardware solution optimized for Gemini's complex workloads.
Beyond internal improvements, Frozen V2 is poised to give Google a significant competitive advantage in the rapidly evolving AI landscape. Rivals such as OpenAI, Microsoft, Meta, and Amazon are also investing heavily in custom AI hardware. OpenAI, for instance, recently launched its custom inference processor, "Jalapeño," while Anthropic is exploring chipmaking partnerships with Samsung. Google's in-house chip development underscores a broader industry trend towards vertical integration in AI.
Google's Commitment to Full-Stack AI Innovation
A Google spokesperson affirmed the company's continuous pursuit of innovation, stating, "Our teams are constantly researching and experimenting with new innovations to deliver maximum performance and efficiency for our users and customers." The spokesperson emphasized Google's "full stack approach," where hardware and software are co-designed from the ground up to ensure optimal integration and performance for real-world AI applications.
The introduction of Frozen V2 alongside Google’s existing Tensor Processing Units (TPUs) would further solidify the company's position in advanced AI hardware, ensuring its AI models remain at the forefront of technological capability and efficiency.