CoreWeave Secures Long-Term Nvidia GPU Contract Through 2029

CoreWeave has locked in a long-term contract for Nvidia A100 GPUs, which will extend through 2029. This deal was announced during the company’s Q2 2026 earnings call, where CEO Michael Intrator emphasized that these machines are being booked at 'full freight.'

In the same earnings report, CoreWeave revealed a revenue of $2.575 billion, marking a 112% increase year-over-year. CFO Nitin Agrawal noted that the A100 contract was secured at an attractive price point, indicating strong profitability expectations for the company.

Despite the A100 chips being launched in 2020, CoreWeave has seen a price increase for older-generation GPUs, reflecting a strong demand for its services. The company also reported a revenue backlog of $104.2 billion, excluding over $25 billion in early Q3 commitments.

CoreWeave’s relationship with Nvidia has evolved beyond a simple buyer-seller dynamic, as Nvidia invested $2 billion in the company in January 2026, ensuring a reliable customer for its chips.

Originally founded in 2017 as an Ethereum mining operation, CoreWeave has successfully pivoted to become a cloud infrastructure provider focused on GPU workloads, capitalizing on the growing demand for AI-related services.

New Developments in Nvidia's Financing Strategy

• Nvidia has partnered with six major financial institutions, including Apollo, BlackRock, and Goldman Sachs, to secure over $500 billion in capital for AI data center projects.
• The initiative aims to treat Nvidia's GPUs as investable assets, similar to real estate or aircraft, emphasizing their economic value.
• Nvidia may provide up to 25% residual-value support on select financing projects to reassure lenders about the future value of the hardware.
• The $500 billion target for capital mobilization is comparable to the annual GDP of Norway, highlighting the scale of this initiative.
• In July 2026, Nvidia was negotiating a $250 billion guarantee for a 10-gigawatt data center project associated with OpenAI, with total costs potentially exceeding $500 billion.
• No standardized framework currently exists for appraising the value of data center contents, complicating the valuation process for institutional investors.