Supermicro has begun shipping NVIDIA Vera Rubin NVL72 racks integrated with its Data Center Building Block Solutions (DCBBS) and DLC-2 direct liquid cooling stack. The offering covers cooling infrastructure from cold plates and manifolds to cooling distribution units (CDUs) and facility-side cooling towers.
Each liquid-cooled NVIDIA Vera Rubin NVL72 rack combines 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs. Eighteen 1U compute trays, each containing four Rubin GPUs and two Vera CPUs, connect through nine sixth-generation NVIDIA NVLink switch trays delivering 216 TB/s of scale-up bandwidth. Each rack provides 20.7 TB of HBM4 memory and up to 54 TB of LPDDR5X.
Supermicro’s cooling portfolio includes 1.8 MW in-row CDUs deployed with N+1 redundancy, in-rack CDUs, Liquid-to-Air sidecar CDUs, rear door heat exchangers, cold plates, manifolds, hose kits and cooling towers. The company said it tests and validates every rack with the full liquid cooling stack before deployment.
The DCBBS Blueprint covers power envelopes from 5 MW to gigawatt scale. One Vera Rubin NVL72 Scalable Unit comprises 16 compute racks with 1,152 NVIDIA Rubin GPUs and 331 TB of HBM4 memory, alongside cooling capacity, power delivery, storage and networking.
“We have spent years building the liquid-cooling stack, the manufacturing capacity, and the deployment teams for exactly this moment,” said Charles Liang, president and CEO, Supermicro. “Our customers can now order a Scalable Unit and receive production-ready systems with end-to-end integration, because we design and build every layer between the cold plate and the cooling tower.”
