NVIDIA and Oracle expand AI cloud partnership as Vera Rubin becomes available on OCI

NVIDIA and Oracle announced an expansion of their Vera Rubin AI infrastructure partnership to multi-gigawatt scale. Gigawatts measure electrical power and describe the scale of computing infrastructure here. The companies say Vera Rubin NVL72 systems and Vera CPUs—the processors that run programs and coordinate computing tasks—are now generally available through Oracle Cloud Infrastructure, or OCI. Customers can access the hardware through Oracle's cloud service.^[1]

The companies also plan to introduce HGX Rubin NVL8 systems. They say the products are intended to support more advanced AI model training, improve inference efficiency—the work a trained model does to answer requests—and expand AI applications that carry out multistep tasks. HGX Rubin NVL8 remains a planned addition.^[1]

Offering the hardware through OCI lets customers rent computing resources, reducing their need to buy and manage complete systems themselves. That gives NVIDIA another route to customers and Oracle more AI computing services to sell. The revenue opportunity depends on how much capacity customers use and how the expanded infrastructure is deployed.^[1]

If the products deliver the companies' stated efficiency goals, AI developers could need fewer computing resources for a given task, lowering usage costs or allowing more work. For Oracle, expanding capacity also requires investment in hardware and supporting infrastructure; the return depends on enough paying demand to use that capacity.^[1]

Impact