Next-generation rack-scale AI system from Supermicro with 72 Rubin Ultra GPUs and 36 Vera CPUs in a single liquid-cooled rack. Designed for frontier-scale training, reasoning, and hyperscale inference. Coming soon.
72 next-generation NVIDIA Rubin Ultra GPUs in a single rack, operating as one unified compute domain. Expected to deliver significant gains in compute density and performance per watt over the GB300 NVL72.
36 NVIDIA Vera CPUs purpose-built for tight integration with the GPU fabric. The successor to Grace, designed to further minimize CPU-GPU bottlenecks in AI workloads.
Next-generation NVLink interconnect providing high-bandwidth, low-latency connectivity across all 72 GPUs. The full rack is expected to operate as a single compute node.
Full rack liquid cooling backed by Supermicro's leadership in direct liquid cooling technology and Data Center Building Block Solutions for end-to-end AI factory deployment.
The NVIDIA Vera Rubin NVL72 is the successor to the GB300 NVL72, combining 72 Rubin Ultra GPUs and 36 Vera CPUs into a single unified rack-scale compute domain. NVIDIA has stated the platform is designed to deliver significant gains in performance per watt and lower cost per token compared to previous generations, particularly for large-scale training and mixture-of-experts models.
The Supermicro SRS-VR-NVL72 is one of the Vera Rubin NVL72 systems that will be available through Arc Compute. Supermicro brings industry-leading direct liquid cooling technology, manufacturing capacity, and end-to-end data center building block solutions to this platform. Our team is actively supporting early infrastructure planning for organizations building toward Rubin-generation rack-scale AI.
Every cluster is designed around your workloads, your facility, and your operational model.
Systems are validated, burned in, and shipped with full deployment support.
Hands-on experience designing and deploying high-density GPU infrastructure.
From planning and procurement through deployment, operations, and scaling.
