Google is reportedly negotiating a multibillion-dollar agreement to supply its custom AI chips to Meta Platforms, a move that would mark a significant strategic shift for Google as it seeks to challenge Nvidia’s dominance in the AI infrastructure market. According to a report from The Information on Monday evening, the proposed deal would see Meta deploying Google’s Tensor Processing Units (TPUs) within its own data centers starting in 2027. Additionally, Meta plans to begin renting TPU capacity directly from Google Cloud as early as next year.
This potential partnership represents a departure from Google’s traditional business model, which has historically restricted access to its TPUs exclusively through cloud rentals on the Google Cloud Platform rather than selling the hardware for outside use. Following the report, Alphabet shares rose between 2.1% and 2.5% in after-hours trading, while Nvidia stock dipped 1.8%. The news arrived shortly after Alphabet stock had already surged more than 6% during the regular session, buoyed by the positive reception of its recently launched Gemini 3 AI model.
Google is actively pitching its TPU technology to a broader range of clients beyond Meta, including high-frequency trading firms and major financial institutions. The company is emphasizing that on-premises deployment of these chips can help organizations meet strict security and compliance requirements.
Currently, Meta relies heavily on Nvidia GPUs to power the AI infrastructure serving its more than 3 billion daily users. Google Cloud executives estimate that expanding TPU adoption could allow the company to capture up to 10% of Nvidia’s annual revenue, translating to billions in potential earnings amidst ongoing global supply constraints for AI computing power.
The deal would serve as a major validation of Google’s decade-long investment in custom silicon. The company recently introduced Ironwood, its seventh-generation TPU, which claims to offer four times the performance of its predecessor and is nearly 30 times more energy-efficient than the first Cloud TPU released in 2018.
Competition in the sector is intensifying, with AI startup Anthropic committing in October to access up to one million Google TPUs in a deal valued at tens of billions of dollars, citing price-performance and efficiency as decisive factors. Google continues to partner with Broadcom for the design and manufacturing of these chips.
