Supermicro introduced a blueprint for an eight-rack NVIDIA Vera Rubin system, detailing modular GPU infrastructure configuration for scale deployment.

On June 22, 2026, Supermicro introduced a blueprint for an eight-rack liquid-cooled system built around NVIDIA's Vera Rubin NVL4 platform, designed to scale from high-performance computing to artificial intelligence workloads. The Data Center Building Block Solutions (DCBBS) unit comprises eight liquid-cooled compute racks housing 288 nodes with up to 1,152 NVIDIA Rubin GPUs and 576 NVIDIA Vera CPUs, operating within a 3.2-megawatt power envelope and capable of scaling from 3.2 megawatts to 1 gigawatt.

Each rack features a 362-kilowatt power envelope and contains 36 NVIDIA Vera Rubin NVL4 nodes. Supermicro's DLC-2 direct-liquid-cooling system uses three in-row cooling-distribution units for each scalable unit, paired with direct-to-chip copper cold plates. The architecture integrates Rubin GPUs, Vera CPUs, ConnectX-9 SuperNICs, and BlueField-4 DPUs, targeting high-performance computing and AI applications including scientific simulation and data processing.

The blueprint represents an architectural framework rather than a confirmed production milestone. Supermicro has not identified a completed shipment, customer, purchase order, revenue figure, or deployed site. Vera Rubin NVL4 systems from global manufacturers, including Supermicro, are expected to become available in the fourth quarter of 2026.

"Scientific discovery has always been driven by the tools available to researchers, and AI has become an essential part of the research process," said Supermicro President and CEO Charles Liang. "With our DCBBS Blueprints for NVIDIA Vera Rubin NVL4, research organizations can confidently deploy HPC and AI infrastructure at any scale."

NVIDIA founder and CEO Jensen Huang characterized the platform as "a new instrument for science — a rack-scale supercomputer that brings simulation, AI and data processing together."