Oracle, NVIDIA expand AI infrastructure partnership on OCI

Traders Union
All News
Business & Wealth
Companies
Oracle, nvidia expand ai partnership

Highlights

Oracle and NVIDIA will expand OCI Superclusters with NVIDIA Vera Rubin NVL72 systems to multi-gigawatt scale, supporting larger AI model workloads.
Oracle will launch NVIDIA HGX Rubin NVL8 servers featuring eight Rubin GPUs, dual 64-core Intel Xeon 6 CPUs, 4TB DDR5 memory, and 6.4 Tbps RDMA networking for efficient, x86-compatible AI deployment.
OpenAI trained and served GPT-6 Astra in Abilene on an OCI Supercluster with over 100,000 NVIDIA Blackwell GPUs, with Vera CPUs entering production to meet agentic AI workload demands.

OCI expansion plan and new platform rollout

As reported by IPRSoftware Media, citing NVIDIA Newsroom, the companies say OCI Superclusters are set to scale with NVIDIA Vera Rubin NVL72 systems to a multi-gigawatt level, extending an existing buildout that already includes NVIDIA GB200 and GB300 NVL72 infrastructure. The expansion is designed to support larger AI model training and serving workloads while letting customers scale CPU and GPU capacity independently.

OCI Superclusters combine NVIDIA NVL72 rack-scale systems, Oracle Acceleron networking, NVIDIA Spectrum-X Ethernet networking with Spectrum-6 switches and BlueField-4 DPUs, plus gigawatt-class power and cooling. Each rack links 72 GPUs through NVLink scale-up networking, while OCI's multiplanar network is built to scale across hundreds of thousands of GPUs.

Oracle also plans to introduce NVIDIA HGX Rubin NVL8, an x86-based server configuration with eight Rubin GPUs connected through NVLink. The platform is set to include two 64-core Intel Xeon 6 CPUs, 4TB of DDR5 memory and 6.4 Tbps of RDMA networking, giving customers a smaller Rubin deployment option for training, fine-tuning and inference while maintaining compatibility with x86-based applications.

AI workload demand and customer implications

Both companies position the rollout around rising demand for agentic AI, where CPU performance becomes more important between model calls for tasks such as tool execution, code testing and data retrieval. OCI Compute with NVIDIA Vera CPUs is intended to reduce GPU idle time by improving single-thread performance and memory bandwidth for those CPU-intensive operations.

OCI Kubernetes Engine lets customers scale CPU and GPU resources separately, and Oracle says Acceleron adds hardware-based isolation for workloads that run untrusted, AI-generated code. The companies argue that this setup can help customers use infrastructure more efficiently as model requirements change across training and inference.

OpenAI is cited as an early user of the infrastructure. In Abilene, the company has trained and served GPT-6 Astra on an OCI Supercluster with more than 100,000 NVIDIA Blackwell GPUs, while executives say they are also bringing Vera CPUs into production on OCI to support growing agentic sandbox workloads at scale.

In our earlier article on the US AI data center boom, we highlighted how soaring buildout plans are increasingly colliding with higher financing costs, construction delays, and tightening power availability. We also noted that only a portion of planned projects is seen as realistically deliverable, with investor confidence and local grid/community constraints becoming key tests for how fast compute capacity can expand.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.