New report claims Meta to spend billions on Google chips

Meta Platforms is reportedly in negotiations with Google to spend billions of dollars on the latter's custom-built artificial intelligence (AI) chips for use in its data centres. The move, if finalised, represents a significant shift in the competitive landscape of the AI hardware market, positioning Google as a serious rival to chip giant Nvidia.

The discussions, initially reported by The Information, involve Meta potentially beginning to rent Google’s Tensor Processing Units (TPUs) via the Google Cloud platform as early as next year. Furthermore, the social media firm is contemplating purchasing and deploying these chips in its own data centres from 2027 onwards. A deal of this scale would mark a major win for Google, given that Meta is one of Nvidia’s largest customers, with planned spending on AI infrastructure reaching tens of billions this year.

This strategic push is part of Google's wider effort to persuade major clients to adopt its TPUs for AI workloads within their own facilities. This represents a substantial departure from Google’s traditional approach of using TPUs almost exclusively for its internal infrastructure. Executives at Google Cloud suggest that expanding TPU adoption could allow the company to capture as much as 10% of Nvidia’s annual data centre revenue.

Demand has surged for custom chips like TPUs, as businesses seek alternatives to Nvidia's expensive and supply-constrained graphics processors. News of the potential deal caused an immediate reaction on financial markets, with Alphabet (Google’s parent company) shares rising, while Nvidia's stock experienced a noticeable decline. Neither Alphabet, Meta, nor Nvidia have issued an official comment on the reports, hence it remains unverified.