SMCI•SMCI•Supermicro’s new DCBBS Blueprint for the NVIDIA Vera Rubin NVL4 platform delivers up to 1,152 GPUs and 576 CPUs per 3.2MW liquid-cooled Scalable Unit, supporting 362 kW per rack with DLC-2 cooling. The end-to-end offering covers facility surveys, system integration, deployment, and support to expedite large-scale HPC and AI clusters.
Supermicro launched its DCBBS Blueprint for the NVIDIA Vera Rubin NVL4 platform, delivering a complete end-to-end solution designed to converge high-precision FP64 simulation and AI computing for scientific research institutions and supercomputing centers.
Each Scalable Unit comprises eight 52U liquid-cooled racks housing 288 NVL4 nodes, up to 1,152 NVIDIA Rubin GPUs and 576 Vera CPUs, with DLC-2 direct liquid cooling providing 362 kW per rack via three in-row CDUs and SMC PG25-A coolant for optimal thermal performance.
The solution lifecycle spans on-site facility surveys assessing power, cooling and floor load, followed by rack-level integration, L10 and L11 testing in Supermicro’s manufacturing facilities. White-glove delivery includes rack placement, power and network cabling, commissioning and optional 4-hour on-site response support for mission-critical environments.
By offering a pre-validated, modular AI and HPC infrastructure, Supermicro aims to reduce time-to-online and total cost of ownership, positioning itself as a provider of choice for research institutions. Rapid deployment capabilities and scalable architecture could drive significant revenue growth in scientific computing markets.
