CoreWeave and Meta Forge $21 Billion AI Compute Pact: A Deep Dive

An expansive long‑term deal reshapes the AI infrastructure race through 2032

On April 9, 2026, CoreWeave and Meta expanded their existing AI cloud partnership with a staggering $21 billion agreement, extending dedicated GPU‑powered capacity through December 2032. This cements Meta’s strategic reliance on specialized AI clouds amid its massive internal AI infrastructure build‑out.

The Deal Unveiled: Scale, Scope, and Significance

On April 9, 2026, CoreWeave announced a $21 billion expansion of its AI cloud computing agreement with Meta Platforms, extending dedicated GPU‑based capacity through December 2032 . This builds atop the earlier $14.2 billion deal signed in September 2025, bringing the total value of contracts between the two companies to approximately $35 billion . The infrastructure under the deal spans multiple data centers and includes some of the first deployments of Nvidia’s next‑generation Vera Rubin GPU platform, enabling Meta to scale both training and inference workloads with high performance and resilience .

Strategic Drivers: Why Meta Is Betting Big on CoreWeave

Meta’s AI ambitions are propelling its infrastructure spending to unprecedented levels. For 2026, the company has budgeted between $115 billion and $135 billion in capital expenditures, with AI compute and data center capacity identified as central cost drivers . CEO Mark Zuckerberg has emphasized that compute capacity remains a key bottleneck to AI progress, making external partnerships like this one essential while Meta scales its internal footprint . By securing long‑term, geographically distributed GPU capacity via CoreWeave, Meta hedges against infrastructure bottlenecks and accelerates access to cutting‑edge hardware—particularly critical as it rolls out new models like Muse Spark and refines its AI Superintelligence Labs operations .