started · updated
NVIDIA and Oracle expand AI infrastructure with Vera Rubin systems
NVIDIA and Oracle have announced plans to expand Oracle Cloud Infrastructure (OCI) Supercluster to a multi-gigawatt scale using NVIDIA Vera Rubin NVL72 systems. The collaboration aims to meet rising demand for AI training, inference, and agentic applications by providing general availability for both NVIDIA Vera Rubin NVL72 and NVIDIA Vera CPUs on OCI.
As part of this expansion, Oracle plans to introduce the NVIDIA HGX Rubin NVL8, an x86-based configuration featuring two 64-core Intel Xeon 6 CPUs and eight Rubin GPUs. This setup is designed to support AI fine-tuning and inference while maintaining compatibility with x86-based applications.
Concurrently, NVIDIA has updated the hardware specifications for its Rubin GPU as production begins. The revised specifications show a reduction in HBM4 memory bandwidth and NVLink scale-up bandwidth compared to preliminary figures, though scale-out networking bandwidth has increased. The Rubin GPU features 288GB of HBM4 memory per unit.
Entities
ASRock Rack · Nvidia · OpenAI · Oracle · Oracle Cloud Infrastructure