Super Micro Computer Inc. (NASDAQ: SMCI) announced Data Center Building Block Solutions blueprints designed for NVIDIA's Vera Rubin NVL72 and HGX Rubin NVL8 platforms. The blueprints are engineered to scale from 5MW to 1GW power envelopes for AI data center deployments.

The DCBBS blueprints include compute, storage, networking, liquid cooling, power distribution and site infrastructure components. Supermicro stated the solutions can accommodate single-tenant or multi-tenant deployments with full facility-side infrastructure.

Key technical specifications include 1,152 NVIDIA Rubin GPUs with 331TB of HBM4 GPU memory per scalable unit. The system features DLC-2 direct liquid cooling technology with cold plates, cooling distribution units, manifolds and cooling towers, along with SMC PG25-A coolant.

The blueprints incorporate 16 compute racks optimized for the NVIDIA platforms, six networking racks supporting NVIDIA Spectrum-X Ethernet or Quantum-X800 InfiniBand, four high-performance storage racks, and two context memory storage platform racks.

"The NVIDIA Vera Rubin NVL72 platform sets a new standard for AI factory performance, and our DCBBS Blueprints give customers a proven, end-to-end path to build at any scale — from 5MW to 1GW," said Charles Liang, president and CEO of Supermicro.

The company stated it provides dedicated teams to manage deployment lifecycle including site surveys, project design, integration and ongoing support. Supermicro claims experience deploying liquid-cooled AI facilities with over 100,000 GPUs.

Deployments are scheduled for the second half of 2026, aligned with NVIDIA Vera Rubin general availability. Supermicro plans to demonstrate the platforms at Computex in Taipei, Taiwan from June 2-6, 2026.