SAN FRANCISCO
US tech giant Meta Platforms made a $21 billion long-term deal with cloud compute provider CoreWeave to scale up its artificial intelligence (AI) infrastructure over the next six years, according to the cloud firm on Thursday.
The partnership will deepen the already extensive relationship between the social media giant and the AI cloud operator, extended through December 2032, the press release showed.
CoreWeave will deploy dedicated cloud capacity across data centers to meet the needs of a strategic industry shift toward “AI inference,” which requires immense compute power to operate and deploy trained AI models.
CoreWeave’s solution for Meta will feature some of the industry’s first installations of chip giant Nvidia’s next-generation Vera Rubin graphics processing unit (GPU) platform.
The deal is a clear sign of the industry’s rising demand for high-performance infrastructure for rapidly evolving AI apps, CoreWeave co-founder and CEO Michael Intrator said.
The $21 billion commitment will supplement a prior $14 billion deal the two firms signed last year amid the social media giant’s rising need for specialized compute power, which reportedly outperforms its in-house data center capabilities.
The Facebook parent company is trying to secure state-of-the-art GPUs and plans to spend up to a whopping $135 billion on its AI scaling in 2026 alone.
The contract will cement CoreWeave’s place in the industry as an infrastructure provider capable of supplying specialized compute capacity that the world’s largest tech companies need to hyperscale their growing AI operations.
CoreWeave said it plans to raise an additional $4.25 billion via debt offerings to fund its data center expansion to meet the demand.
