CoreWeave disclosed in an SEC filing that Meta has signed a new order form under their existing Master Services Agreement (MSA), committing up to $14.2 billion for cloud computing capacity through December 14, 2031. The deal, dated September 25, 2025, allows Meta to expand its capacity commitment into 2032 for additional services.
The MSA, originally executed in December 2023, governs how Meta accesses CoreWeave’s GPU cloud infrastructure through reserved capacity orders. Both companies retain the right to amend, expand, or terminate orders under standard provisions for cause. The agreement also includes customary terms covering warranties, indemnification, and liability limits.
CoreWeave, which has rapidly scaled its AI-focused cloud infrastructure business, noted that the size and scope of this order makes the MSA a material definitive agreement. This filing follows a series of large-scale cloud capacity deals between CoreWeave and major hyperscalers and AI firms, reinforcing the company’s role as a top supplier of GPU-accelerated compute at scale.
- Meta committed up to $14.2B in cloud capacity through December 2031
- Option to extend and expand commitment into 2032
- Agreement structured under the December 2023 MSA between CoreWeave and Meta
- MSA includes termination-for-cause and liability limitation provisions
- Deal underscores growing demand for CoreWeave’s GPU cloud infrastructure
“We continue to scale our infrastructure to meet the demands of the world’s largest AI innovators,” CoreWeave said in the filing.
🌐 Analysis: This deal highlights Meta’s long-term commitment to GPU-based cloud resources as it competes in the generative AI and large-scale model training market. For CoreWeave, it provides a multi-year revenue stream comparable in scale to earlier agreements with Microsoft and NVIDIA-backed ventures, cementing its position in the AI cloud ecosystem. Rivals such as Oracle, AWS, and Google Cloud are pursuing similar multi-billion-dollar AI infrastructure commitments with hyperscale clients.
