Next-generation 2U liquid-cooled server from Supermicro with 8 NVIDIA Rubin GPUs and NVLink 6. Supports NVIDIA Vera CPUs or x86 processors. Designed for AI training, inference, and reasoning workloads. Coming soon.
Eight next-generation NVIDIA Rubin GPUs with up to 288 GB HBM4 per GPU and approximately 2.3 TB total GPU memory. Connected via NVLink 6 at 28.8 TB/s total bandwidth.
Supports NVIDIA Vera CPUs (88 Arm Olympus cores) or next-generation x86 processors. Choose the CPU platform that fits your workload and ecosystem requirements.
Sixth-generation NVLink delivering 28.8 TB/s total GPU-to-GPU bandwidth across all 8 GPUs. Purpose-built for the distributed compute demands of reasoning and training workloads.
Built on Supermicro's Data Center Building Block Solutions with direct liquid cooling and an optional Liquid-to-Air sidecar CDU for data centers without facility-side liquid cooling infrastructure.
The HGX Rubin NVL8 is the successor to HGX Blackwell, delivering up to 5.5x more NVFP4 FLOPS and up to 3.5x higher inference performance compared to the previous generation. The platform is purpose-built for the emerging era of reasoning and agentic AI, where models require significantly more compute per query.
The Supermicro SYS-RB-NVL8 is one of the HGX Rubin NVL8 systems that will be available through Arc Compute. Supermicro brings industry-leading direct liquid cooling technology, flexible CPU support (Vera and x86), and a Liquid-to-Air sidecar CDU option that enables liquid-cooled Rubin deployments in air-cooled facilities. Our team is actively supporting early infrastructure planning for organizations building toward Rubin-generation 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.
