Nvidia and Oracle are taking their AI partnership to a scale usually reserved for power plants. The two companies announced on October 27, 2026 that Oracle Cloud Infrastructure (OCI) Superclusters will grow to multi-gigawatt capacity using Nvidia’s Vera Rubin platform.
The announcement goes beyond a promise. Vera Rubin NVL72 racks and Nvidia’s Vera CPUs are now available on OCI, so customers can rent the hardware today rather than read about it in a roadmap.
What Oracle is actually getting
Vera Rubin is the successor to Nvidia’s Grace Blackwell generation.
The flagship configuration is the NVL72 rack. It packs dozens of GPUs into a single cabinet and stitches them together with NVLink 6, Nvidia’s latest chip-to-chip interconnect.
Memory is the other headline spec. Each Vera Rubin GPU uses HBM4 memory supporting up to 288 GB and 22 TB/s of bandwidth.
Oracle will wrap these racks in advanced Ethernet networking and Nvidia’s BlueField-4 DPUs. A DPU, or data processing unit, handles networking, storage and security chores so the GPUs can stay focused on AI work.
The deal also brings HGX Rubin NVL8 systems into the mix. These smaller eight-GPU configurations are aimed at frontier model training, inference and agentic AI workloads.
The performance claims
Nvidia is promising big generational jumps. Vera Rubin is designed to deliver up to 30 times higher inference throughput per watt compared to Blackwell.
On the training side, Vera Rubin is projected to deliver four times the performance of the Blackwell generation. These are Nvidia’s figures, and real-world results will depend on workloads and configurations.
The OCI setup allows CPU and GPU resources to scale independently. Customers no longer have to buy processing power in fixed bundles.
Why the timing matters
This expansion builds on Oracle’s earlier deployments of Nvidia Blackwell NVL72 systems. Production shipments of Vera Rubin began ramping in fall 2026.
Nvidia has put a striking number on the opportunity. The company estimates a $40 billion revenue opportunity per gigawatt for the Vera Rubin platform.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.