Google Intensifies AI Chip Battle as Meta Considers Multibillion-Dollar TPU Deal

Google is ramping up its challenge to Nvidia in the accelerating AI chip race, with a new strategy that could dramatically expand the reach of its custom tensor processing units (TPUs). According to a report from The Information, Google is now promoting its AI chips for installation directly inside customers’ data centers—a major shift from its long-standing approach of offering TPU access exclusively through Google Cloud.

One of the most significant potential customers is Meta Platforms Inc., the parent company of Facebook and Instagram. Meta is reportedly in talks to invest billions to deploy Google’s TPUs across its global data centers starting in 2027. The company may also begin renting TPU compute capacity from Google Cloud as early as next year. Meta currently relies heavily on Nvidia GPUs to power its massive AI infrastructure, including its growing portfolio of generative AI models.

News of the discussions immediately moved markets. Alphabet shares climbed 2.1% in after-hours trading, while Nvidia’s stock slipped 1.8%, reflecting investor reaction to a possible shift in the competitive landscape. If finalized, the Meta deal would represent a major endorsement of Google’s hardware ambitions and strengthen its position in the AI computing market, where Nvidia has long dominated.

Google reportedly believes that wider TPU adoption could help the company capture as much as 10% of Nvidia’s annual revenue, representing billions in potential gains. The company has been pitching TPUs to a range of clients, including major financial institutions and high-frequency trading firms, emphasizing that on-premises deployment can help organizations meet strict data security and compliance standards.

As global demand for AI compute continues to surge and supply constraints persist, Google’s push to place TPUs directly into customer facilities underscores a more aggressive phase in the AI chip wars. With major players like Meta evaluating alternatives to Nvidia, the competitive landscape for AI hardware is poised for significant change.