CoreWeave disclosed in a new SEC filing that it has entered into a $6.3 billion agreement with NVIDIA under its existing Master Services Agreement (MSA). The order, dated September 9, 2025, covers the sale of reserved cloud computing capacity to CoreWeave customers and grants NVIDIA access to purchase any unused capacity through April 13, 2032.
The arrangement ensures that any residual, unsold datacenter capacity from CoreWeave will be purchased by NVIDIA, providing CoreWeave with guaranteed utilization while strengthening NVIDIA’s access to cloud compute resources. The agreement builds on a partnership that dates back to April 2023, when the two companies first signed the MSA. NVIDIA also remains a GPU supplier and a stockholder in CoreWeave.
The MSA includes standard terms covering breach, insolvency, indemnification, and liability limitations. Either party can terminate the agreement with 30 days’ notice in the event of a breach, or if the other enters bankruptcy or insolvency proceedings not resolved within 90 days. The agreement will remain in effect until 2032 unless terminated earlier.
Tbps
- Order form under the MSA valued at $6.3 billion.
- Covers reserved CoreWeave cloud capacity with NVIDIA obligated to purchase residual capacity.
- Runs through April 13, 2032, subject to delivery and service requirements.
- Builds on the original MSA signed in April 2023.
- NVIDIA is both a GPU supplier and stockholder in CoreWeave.
🌐 Analysis: This deal locks in long-term demand for CoreWeave’s AI-optimized cloud capacity while giving NVIDIA priority access to compute resources to support its expanding ecosystem of AI developers. The structure provides CoreWeave with revenue stability in a volatile market and positions NVIDIA to hedge against supply shortages in hyperscale AI infrastructure. It echoes similar strategies seen in cloud partnerships between hyperscalers and chip suppliers, underscoring how tightly compute supply chains are being integrated.
🌐 We’re tracking the latest developments in AI infrastructure and hyperscale cloud. Follow our ongoing coverage at: https://convergedigest.com/category/ai-infrastructure/