{"slug": "72-gpu-nvidia-vera-rubin-nvl72-racks-anchor-ocis-planned-multi-gigawatt", "title": "72-GPU NVIDIA Vera Rubin NVL72 Racks Anchor OCI’s Planned Multi-Gigawatt Expansion", "summary": "NVIDIA Vera Rubin NVL72 systems and NVIDIA Vera CPUs are now generally available on Oracle Cloud Infrastructure, and NVIDIA and Oracle announced at Oracle AI World a plan to expand OCI Superclusters with Vera Rubin NVL72 racks to multi-gigawatt scale. Each Vera Rubin NVL72 rack connects 72 GPUs through NVLink, and OCI Superclusters pair those racks with Oracle Acceleron networking, NVIDIA Spectrum-X Ethernet with Spectrum-6 switches and BlueField-4 DPUs; Oracle says the multiplanar network scales across hundreds of thousands of GPUs. The companies cite OpenAI, which trained and served GPT-6 Astra in Abilene on an OCI Supercluster with more than 100,000 NVIDIA Blackwell GPUs, and OpenAI is an early-access customer for Vera CPUs, which target the tool calls, code execution and data retrieval agents run between model calls.", "body_md": "NVIDIA and Oracle plan to expand Oracle Cloud Infrastructure (OCI) Superclusters with [NVIDIA Vera Rubin NVL72](https://www.storagereview.com/news/nvidia-gtc-2026-rubin-gpus-groq-lpus-vera-cpus-and-what-nvidia-is-building-for-trillion-parameter-inference) systems to multi-gigawatt scale, and both Vera Rubin NVL72 and NVIDIA Vera CPUs are now generally available on OCI. The companies announced the pair at Oracle AI World. The expansion is a plan; the general availability is the part customers can buy now, aimed at the GPU-heavy work of training and serving frontier models and the CPU-heavy work agents do between model calls, with CPU and GPU capacity scaled independently.\n\nThe proof point the companies offer is OpenAI, which has trained and served GPT-6 Astra in Abilene on an OCI Supercluster with more than 100,000 NVIDIA Blackwell GPUs. “As our frontier models advance, we need infrastructure that can keep pace,” said Sreekanth Pothanis, engineering leader, compute infra at OpenAI. “NVIDIA Vera Rubin on OCI gives us the computing performance and scale to accelerate model training and serve increasingly demanding AI workloads.”\n\n## NVL72 Racks and Supercluster Networking\n\nEach Vera Rubin NVL72 rack connects 72 GPUs through NVLink scale-up networking. OCI Superclusters combine those racks with [Oracle Acceleron](https://www.storagereview.com/news/oracle-acceleron-front-end-rdma-occl-plugin-multiplanar-failover-3x-nvme-block-storage) networking, NVIDIA Spectrum-X Ethernet with [Spectrum-6](https://www.storagereview.com/news/nvidia-cisco-usfp-optics-400gbps-per-lane-ocp-2026) switches and BlueField-4 DPUs, and gigawatt-class power and cooling. Oracle says the multiplanar network scales across hundreds of thousands of GPUs. The Vera Rubin build-out extends the GB200 and GB300 NVL72 systems already on OCI, and the companies point to the common CUDA foundation as the reason models and workloads carry across generations.\n\n## HGX Rubin NVL8 Brings Eight GPUs to x86 Servers\n\nOCI also plans to add NVIDIA [HGX Rubin NVL8](https://www.storagereview.com/news/supermicro-expands-liquid-cooling-for-nvidia-vera-rubin-nvl72-and-hgx-rubin-nvl8-platforms), an x86 server with eight Rubin GPUs connected through NVLink. The planned configuration carries two 64-core Intel Xeon 6 CPUs, 4TB of DDR5 memory, and 6.4 Tbps of RDMA networking, and Oracle positions it for training, fine-tuning, and inference where a customer wants Rubin GPUs in a smaller footprint without giving up x86 application compatibility. Neither company gave a date for it.\n\n## Vera CPUs Support Agent Tool Execution\n\nThe Vera CPU offering targets the work an agent does between model calls: tool calls, code execution, and data retrieval. Those steps run on CPUs, and when they run slowly the GPUs sit waiting. NVIDIA and Oracle say Vera’s single-thread performance speeds up individual tool calls and code execution while its memory bandwidth keeps cores fed as agents run concurrent tasks, which together cut GPU idle time. Neither company gives performance figures or test conditions for the claim.\n\nOCI Kubernetes Engine lets customers scale CPU and GPU pools independently, and Oracle Acceleron provides hardware-based isolation for workloads running untrusted, AI-generated code. OpenAI is among the early-access customers. “NVIDIA Vera CPUs on OCI deliver the world-class loaded performance needed to execute code and run tools across our rapidly growing agentic sandbox workloads,” said Uday Ruddarraju, chief technology officer of compute at OpenAI. “We are bringing Vera into production on OCI to support these workloads at scale.” The scale of that deployment was not given.\n\n“AI factories must be productive, durable and fungible,” said Ian Buck, vice president of hyperscale and high-performance computing at NVIDIA, describing the goal as infrastructure that stays useful through software advances and serves different models and workloads as demand changes. Mahesh Thiagarajan, president of Oracle Cloud Infrastructure, framed the pairing of Vera Rubin GPUs and Vera CPUs with Acceleron networking as a way to keep GPUs productive and secure as customer needs grow.", "url": "https://wpnews.pro/news/72-gpu-nvidia-vera-rubin-nvl72-racks-anchor-ocis-planned-multi-gigawatt", "canonical_source": "https://www.storagereview.com/news/nvidia-vera-rubin-nvl72-vera-cpus-generally-available-oci-multi-gigawatt-superclusters", "published_at": "2026-10-29 15:32:37+00:00", "updated_at": "2026-10-29 15:47:33.412331+00:00", "lang": "en", "topics": ["ai-infrastructure", "ai-chips", "ai-agents", "ai-products"], "entities": ["NVIDIA", "Oracle", "Oracle Cloud Infrastructure", "Vera Rubin NVL72", "NVIDIA Vera CPU", "OpenAI", "GPT-6 Astra", "Ian Buck"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/72-gpu-nvidia-vera-rubin-nvl72-racks-anchor-ocis-planned-multi-gigawatt", "markdown": "https://wpnews.pro/news/72-gpu-nvidia-vera-rubin-nvl72-racks-anchor-ocis-planned-multi-gigawatt.md", "text": "https://wpnews.pro/news/72-gpu-nvidia-vera-rubin-nvl72-racks-anchor-ocis-planned-multi-gigawatt.txt", "jsonld": "https://wpnews.pro/news/72-gpu-nvidia-vera-rubin-nvl72-racks-anchor-ocis-planned-multi-gigawatt.jsonld"}}