OpenAI has signed a new $6.5 billion contract with CoreWeave, expanding their existing partnership to a combined value of $22.4 billion in 2025. This latest deal marks the third expansion this year, following contracts of $11.9 billion in March and $4 billion in May.
The contract is designed to help build out OpenAI’s “Stargate” infrastructure project, a multi-year initiative targeting 10 gigawatts of compute capacity across global sites. With this addition, committed capacity between OpenAI and CoreWeave now approaches 7 gigawatts, making CoreWeave one of OpenAI’s most important suppliers of AI compute.
Nvidia’s Role in the Partnership
Nvidia is deeply embedded in this arrangement. CoreWeave’s data centers rely heavily on Nvidia GPUs for AI workloads, and Nvidia itself owns more than 5% of CoreWeave. In addition, Nvidia and CoreWeave signed a $6.3B agreement in September under which Nvidia committed to purchase any unused CoreWeave capacity through 2032. This interlocking set of agreements has drawn attention from analysts and regulators.
Where CoreWeave Fits
Founded in 2017 and originally focused on cryptocurrency mining, CoreWeave pivoted to GPU cloud infrastructure and has rapidly grown into one of the largest specialized AI cloud providers in the United States. The company operates data centers across North America and Europe and has become a preferred partner for organizations requiring dense GPU clusters.
Concerns and Risks
While the deal strengthens OpenAI’s compute pipeline, it raises several questions:
• Circular financing: Nvidia invests in both OpenAI and CoreWeave, while CoreWeave buys Nvidia hardware and has a guaranteed offtake agreement with Nvidia. Critics argue this loop could blur the line between strategic investment and closed-ecosystem financing.
• Regulatory scrutiny: With a handful of companies (OpenAI, CoreWeave, Nvidia, Microsoft, Oracle) increasingly dominating AI infrastructure, regulators are likely to examine whether such concentration distorts competition.
• Execution risk: Scaling to multi-gigawatt facilities requires massive power, cooling, and logistical coordination. Even slight delays could have knock-on effects for OpenAI’s infrastructure roadmap.
Broader Implications
The deal highlights the pace of AI infrastructure expansion in 2025. Only months ago, billion-dollar contracts in cloud compute were notable outliers. Now, multibillion-dollar agreements are being signed back-to-back. The CoreWeave–OpenAI partnership underscores both the enormous capital flowing into AI and the increasing reliance on tightly coupled ecosystems of suppliers and investors.
For the wider technology sector, this means:
- Hardware supply chains may prioritize hyperscale AI projects over other industries.
- Cloud competition could narrow, as smaller providers struggle to match the scale of CoreWeave’s commitments.
- Policy and oversight will play a growing role in determining how concentrated this infrastructure can become.
Final Words
The $6.5B contract between OpenAI and CoreWeave adds another layer to one of the most significant infrastructure partnerships in AI. With Nvidia intertwined as both supplier and shareholder, the deal cements CoreWeave’s position as a backbone provider of compute capacity for OpenAI’s Stargate project. It also raises fundamental questions about concentration, financing structures, and how far and fast AI infrastructure can scale.
