Meta doubles down on CoreWeave
Meta expanded its CoreWeave contract by another $21 billion through 2032, bringing its committed AI cloud spend to $35 billion and targeting early access to Nvidia Vera Rubin hardware. That deal reinforces the idea that long‑term chip and datacentre agreements are becoming strategic moats for large AI builders. (thenextweb.com)
Meta just agreed to spend another $21 billion with CoreWeave, a company that rents out artificial intelligence computing power, and that pushes Meta’s total committed spend with CoreWeave to about $35.2 billion through December 2032. The new contract starts in 2027 and sits on top of a previous $14.2 billion agreement the two companies signed in September 2025. (coreweave.com) (cnbc.com) CoreWeave is not a consumer app or a chip designer. It is a cloud company whose data centers are packed with Nvidia graphics processing units, the specialized chips that train and run large artificial intelligence models. (cnbc.com) This deal is about renting capacity, not just buying hardware. Meta is building its own data centers at the same time, but CoreWeave chief executive Michael Intrator told CNBC that big customers still buy from CoreWeave because they want extra capacity and lower execution risk. (cnbc.com) CoreWeave said the new capacity will be spread across multiple locations and will include some of the first deployments of Nvidia’s Vera Rubin platform. Nvidia introduced Vera Rubin on March 16, 2026 as a rack-scale system that combines a new central processor, a new graphics processor, networking, and storage into one tightly linked artificial intelligence machine. (coreweave.com) (nvidianews.nvidia.com) Meta is aiming a lot of this rented capacity at inference workloads, which is the part where a trained model answers questions, writes code, or generates images for real users. Training is like teaching the model in a classroom once, while inference is like serving millions of customers at a checkout counter every day. (coreweave.com) (nvidianews.nvidia.com) The timing fits Meta’s wider spending surge. In January 2026, Meta said it expected $115 billion to $135 billion in capital expenditures for the year, up from $72.22 billion in 2025, and said most of the increase would come from infrastructure costs including third-party cloud spend. (datacenterdynamics.com) Meta has also said it is still capacity-constrained, which means its teams want more computing power than the company can currently supply internally. That helps explain why a company rich enough to build its own campuses is still locking in outside cloud contracts that run for six years or more. (datacenterdynamics.com) (cnbc.com) For CoreWeave, the Meta contract is also a balance-sheet story. On the same day it announced the expanded deal, the company said it would raise $3 billion in fresh debt, and investors immediately treated the Meta commitment as proof that future demand is real enough to finance against. (cnbc.com) (finance.yahoo.com) That is why these agreements now look less like ordinary cloud contracts and more like supply treaties. In artificial intelligence, the scarce thing is no longer just talent or software code, but guaranteed access to chips, power, and buildings years before the rest of the market needs them. (coreweave.com) (datacenterdynamics.com)
