CoreWeave, a fast-growing AI cloud infrastructure provider, has signed a $6.3 billion guaranteed order with Nvidia, securing one of the largest cloud capacity agreements in the sector to date. Under the revised terms, Nvidia commits to purchasing any unused compute capacity from CoreWeave through April 2032, effectively underwriting the provider’s long-term data center buildout.
The deal cements CoreWeave’s status as a critical enabler of Nvidia’s GPU-driven AI ecosystem while reducing exposure to demand fluctuations from its existing enterprise clients.
A strategic backstop for AI compute scaling
The agreement builds on a previous contract signed in April 2023 and adds a new layer of financial stability to CoreWeave’s aggressive expansion. Analysts at Barclays called the deal a “backstop mechanism,” enabling CoreWeave to scale capacity confidently without the risk of underutilisation — especially as it ramps up beyond anchor clients like Microsoft and OpenAI.
CoreWeave already operates next-gen GPU data centers across the U.S. and Europe, delivering high-performance infrastructure tailored for training and deploying large AI models.
Surging demand, surging costs
In Q2 2025, CoreWeave reported a spike in cloud usage due to growing AI adoption but also revealed that its operating expenses had nearly quadrupled to $1.19 billion. The Nvidia order provides critical assurance to investors and lenders, offering guaranteed revenue as capital spending accelerates.
This is not the first high-value commitment CoreWeave has secured. In March, OpenAI signed a 5-year deal worth $11.9 billion and a separate agreement to pay up to $4 billion through 2029 — positioning CoreWeave as one of the most financially visible players in the AI infrastructure space.
What it means for the AI ecosystem
The Nvidia-CoreWeave deal reflects how foundational infrastructure providers and AI chipmakers are interlocking their roadmaps. As workloads grow more compute-intensive, long-term capacity planning — underpinned by binding commercial agreements — will be essential to meet enterprise expectations for performance, scalability, and availability.
For Nvidia, the arrangement supports GPU sales and broader ecosystem control. For CoreWeave, it means accelerated credibility, resource deployment, and resilience in an increasingly competitive cloud landscape.
