Supermicro begins shipping NVIDIA Vera Rubin NVL72 racks

Super Micro Computer, Inc. has announced it is now shipping NVIDIA Vera Rubin NVL72 racks. These systems are integrated with Supermicro’s Data Center Building Block Solutions (DCBBS) and its DLC-2 direct liquid cooling stack.

The NVIDIA Vera Rubin platform utilizes direct liquid cooling to achieve higher AI throughput per watt than air-cooling methods can provide. Because high-speed AI operations generate significant heat, the system uses a fluid distribution loop involving cold plates, manifolds, and cooling towers to manage thermal loads.

Supermicro’s infrastructure is designed to scale from 5 MW to gigawatt levels. A single scalable unit can include 1,152 GPUs and 331 TB of HBM4. The company provides end-to-end integration, covering everything from mechanical engineering and large-scale manufacturing to on-site deployment and support.