Super Micro Computer ships Vera Rubin racks featuring 1,152 GPUs

Summary

Supermicro Computer, Inc. (SMCI) is now shipping racks featuring NVIDIA's Vera Rubin GPUs, designed for large-scale AI applications. Each rack includes 72 GPUs and 36 Vera CPUs, with a total of 1,152 Rubin GPUs and advanced cooling capabilities, reflecting the increasing demand for efficient thermal management in AI data centers. The cooling systems integrate cutting-edge liquid cooling technologies, which are essential in supporting the high power and thermal requirements of modern GPU configurations.

Tokens

$SMCI$NVDA

Analysis

DCBBS: DCBBS is a Super Micro direct cold plate cooling solution integrated into its rack offerings. It now extends coverage to the full NVL72 stack for Vera Rubin deployments alongside the DLC-2 system. DLC-2: DLC-2 is Super Micro's second-generation direct liquid cooling technology used in high-density server racks. The solution supports end-to-end cooling from cold plate to cooling tower for the Vera Rubin NVL72 configurations being shipped. Vera Rubin: Vera Rubin is NVIDIA's upcoming GPU architecture following Blackwell, incorporating advanced Rubin GPUs paired with Vera CPUs for next-generation AI and accelerated computing workloads. The architecture is being deployed in scalable rack units that Super Micro is now shipping to data center customers. NVIDIA Corporation: NVIDIA Corporation develops graphics processing units, AI accelerators, and related hardware architectures for data centers and high-performance computing. Its Vera Rubin platform, including Rubin GPUs and Vera CPUs, forms the core of the NVL72 rack systems now being shipped by partners in integrated configurations. Super Micro Computer, Inc.: Super Micro Computer, Inc. designs and manufactures high-performance servers, storage systems, and rack solutions optimized for demanding compute environments. The company is shipping complete Vera Rubin-based rack systems to customers, covering the full stack from cold plate cooling through to the tower with its DCBBS and DLC-2 solutions. AI Infrastructure: NVIDIA's successive GPU architectures like Rubin continue to drive specialized rack-level solutions from server partners for large-scale AI training and inference clusters. Data Center Cooling: Advanced liquid cooling technologies are increasingly required to manage the thermal demands of high-power GPU racks in modern AI data centers.

Categories

hyperliquidtech
View Original Tweet