CoreWeave and Meta Platforms Inc. have expanded their long term artificial intelligence (AI) agreement for about US$21 billion (A$29.7 billion).
CoreWeave will provide AI cloud capacity to Meta until December 2032 under the deal, which is additional to a $14.2 billion similar agreement signed in September 2025.
CoreWeave said the capacity would be deployed across multiple locations and include some of the initial deployments of NVIDIA’s Vera Rubin AI computing platform.
It said this “distributed approach" was designed to optimise performance, resilience, and scalability for the AI operations of Meta, which owns social media sites like Facebook and Instagram and data centres and develops AI models, including Llama.
The new agreement was a clear signal of the industry’s accelerating demand for high-performance infrastructure capable of supporting increasingly complex, large-scale AI workloads.
“This is another example that leading companies are choosing CoreWeave’s AI cloud to run their most demanding workloads,” CoreWeave Co-founder, Chief Executive Officer and Chairman Michael Intrator said in a media release.
Nvidia builds the graphics processing units (AI chips), which CoreWeave uses to build cloud infrastructure that is used by companies like Meta.
Meta intends to spend up to $135 billion this year as it rapidly expands its compute capacity to power the development and deployment of its large language models, using the hardware and cloud resources of companies like CoreWeave.
CoreWeave plans up to $35 billion of capital expenditure this year, leveraging its close relationship with Nvidia, to be a major supplier of specialised AI chips like Nvidia's Vera Rubin chips, which are twice as fast as the current Blackwell platform.
CoreWeave (NASDAQ: CRWV) shares closed $3.10 (3.49%) higher at $92.00 on Thursday (Friday AEST), capitalising the company at $48.36 billion.
