Meta has signed a multi-year deal to expand its data centers with millions of Nvidia’s Grace and Vera CPUs and Blackwell and Rubin GPUs. While Meta has long been using Nvidia’s hardware for its AI products, the deal “represents the first large-scale Nvidia Grace-only deployment,” which Nvidia says will deliver “significant performance-improvements per-watt.” [Meta’s] Data Centers.” The deal also includes plans to add Nvidia’s next-generation Vera CPUs to Meta’s data centers in 2027.
Meta is also working on its own in-house chips to run AI models, but according to financial TimesIt has faced “technical challenges and rollout delays” with its chip strategy. Nvidia is also dealing with concerns about depreciation and chip-back loans used to finance the AI buildout, as well as pressure from competition. CNBC notes that Nvidia’s stock fell four percent after a November report about Meta considering using Google’s Tensor chips for AI, and late last year, AMD announced chip arrangements with both OpenAI and Oracle.
Nvidia and Meta did not disclose how much the deal cost, but Meta, Microsoft, Google and Amazon’s AI spending this year is estimated to cost more than the entire Apollo space program.
<a href