Nvidia and CoreWeave expand partnership to build “AI factories” and adopt new platforms

Nvidia and CoreWeave said Monday they are expanding their long-running partnership to accelerate the buildout of large AI-focused data centers—described by the companies as “AI factories”—targeting more than 5 gigawatts of capacity by 2030.

The announcement reflects a widening definition of AI infrastructure. It is not only about GPUs, but also CPUs, networking, storage, orchestration software, and the systems integration required to run large training and inference workloads reliably at scale.

As part of the expanded relationship, CoreWeave will adopt Nvidia’s CPU and storage platforms and deploy multiple Nvidia generations over time. The plan is designed to ensure that CoreWeave’s cloud platform remains aligned with Nvidia’s roadmap as customers demand faster deployment cycles and tighter hardware-software integration.

Both companies positioned the partnership as a response to rapidly growing global compute demand. CoreWeave has been a notable AI-focused cloud provider serving model builders and large technology firms that need dedicated capacity, while Nvidia continues to push deeper into data-center platforms that bundle chips with networking and systems components.

The collaboration also signals an industry shift toward standardized, reference-architecture approaches, where infrastructure providers can roll out validated stacks more quickly. In practical terms, that can shorten the path from procurement to production—an advantage in a market where power availability, physical space, and supply chain lead times increasingly determine who can scale AI services fastest.