AI infrastructure company CoreWeave, Inc. has announced an expanded agreement with OpenAI to supply data center capacity that'll power the training of OpenAI’s “most advanced next-generation models. This latest deal expansion—$6.5 billion—builds on two earlier contracts signed this year—an $11.9 billion pact in March and a $4 billion agreement in May, bringing the total contract value to $22.4 billion.
CoreWeave, which brands itself as the “AI Hyperscaler” that specializes in cloud platforms optimized for demanding AI workload, has become a central player in powering OpenAI’s next-generation models. The two companies latest deal underscores the immense demand for AI computing power, particularly the kind of large-scale, GPU-driven infrastructure required to train advanced systems.
CoreWeave is part of a new wave of so-called “neocloud” providers—firms that specialize in leasing high-end AI chips to customers. Its main rivals include Nebius and Nscale. Just this week, Nscale revealed it had secured $1.1 billion in fresh funding after striking a partnership in the UK with Nvidia and OpenAI.
The rapid growth of neocloud firms, along with the flood of investment pouring into them, highlights the intense demand for AI infrastructure. Major tech players continue to compete fiercely to develop and operate the most advanced AI platforms that help in keeping the need for cutting-edge computing capacity at an all-time high.
OpenAI’s VP of Infrastructure, Peter Hoeschele, commented that CoreWeave “has become an important partner in OpenAI’s infrastructure platform,” delivering compute “at unmatched speed and scale.” Michael Intrator, CoreWeave’s CEO and cofounder, said the expansion “affirms the trust that world-leading innovators have in CoreWeave’s ability to power the most demanding inference and training workloads at an unmatched pace.”
This is the third major deal between the two firms in 2025 alone, and the new OpenAI commitment helps diversify CoreWeave’s business away from Microsoft Corp., which has historically been its largest customer. In the quarter ending June, Microsoft accounted for about 71% of CoreWeave’s total revenue.
Earlier in the year, OpenAI struck an agreement with Oracle Corp. to secure 4.5 gigawatts of computing capacity—an amount comparable to the energy produced by four nuclear power plants. The company also continues to be one of the largest users of Microsoft’s Azure cloud services.
However, these OpenAI deals also follow the steps of other infrastructure deals in the AI ecosystem. For instance, CoreWeave recently sealed a $6.3 billion agreement with Nvidia, in which Nvidia has pledged to purchase any unsold cloud capacity from CoreWeave through 2032. Meanwhile, Nvidia itself is reported to be negotiating an up-to $100 billion investment in OpenAI to support large-scale AI infrastructure development.
