I N T E L L I G E N T T E C H N O L O G Y COOLING
Supermicro announces expanded rack-scale manufacturing capacity for liquid-cooled AI solutions
Supermicro has announced expansions in manufacturing capacity and liquid-cooling capabilities, in collaboration with NVIDIA, to enable first-to-market delivery of data centre-scale solutions optimised for the NVIDIA Vera Rubin and Rubin platforms. Leveraging accelerated development and collaboration with NVIDIA, Supermicro is uniquely positioned to rapidly deploy the flagship NVIDIA Vera Rubin NVL72 and NVIDIA HGX Rubin NVL8 systems. Supermicro’ s proven Data Centre Building Block Solutions( DCBBS) approach delivers streamlined production, extensive customisation options and faster time-to-deployment, giving customers a decisive competitive edge in next-generation AI infrastructure.
“ Supermicro’ s long-standing partnership with NVIDIA and our agile building block solutions enable us to bring the most advanced AI platforms to market faster than others,” said Charles Liang, President and CEO of Supermicro.“ With expanded manufacturing and industry-leading liquid-cooling expertise, we’ re empowering hyperscalers and enterprises to deploy the NVIDIA Vera Rubin and Rubin platforms infrastructure at scale with unmatched speed, efficiency and reliability.”
Flagship products:
• NVIDIA Vera Rubin NVL72 SuperCluster: The premier rack-scale system unifies 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs, NVIDIA ConnectX-9 SuperNICs and NVIDIA BlueField-4 DPUs into a coherent platform with NVIDIA NVLink 6 and scales out with NVIDIA Quantum-X800 InfiniBand and NVIDIA Spectrum-X Ethernet
Supermicro Vera Rubin Rack
to power the AI industrial revolution. It delivers 3.6 exaflops NVFP4 performance, 1.4 PB / s HBM4 bandwidth and 75 TB of fast memory. Built on the third-generation NVIDIA MGX rack architecture for superior serviceability, reliability and availability, Supermicro’ s implementation incorporates an enhanced data centre-scale liquid-cooling technology stack with in-row Coolant Distribution Units( CDUs). This enables scalable warm-water cooling operation that minimises energy consumption and water usage while maximising density and efficiency.
• 2U liquid-cooled NVIDIA HGX Rubin NVL8 systems: This compact 8-GPU system is optimised for AI and HPC workloads, delivering breakthrough performance and efficiency to enterprises for intelligence at scale. It provides 400 petaflops NVFP4, 176 TB / s HBM4 bandwidth, 28.8 TB / s NVLink bandwidth and 1600 Gb / s NVIDIA ConnectX-9 networking SuperNICs. Supermicro delivers rack-scale design with maximum deployment flexibility and configuration options supporting flagship x86 CPUs such as nextgeneration Intel Xeon or AMD EPYC processors. Available options include a high-density 2U busbar design for optimal rack integration featuring Supermicro’ s industry-leading advanced Direct Liquid Cooling( DLC) technology.
Key NVIDIA Vera Rubin Platform features include:
• NVIDIA NVLink 6: High-speed interconnects enabling unprecedented GPU-to-GPU and CPU-to-GPU communication for the training and inference of massive mixture-of-experts models.
• NVIDIA Vera CPU: NVIDIA-designed custom Arm cores delivering 2x performance over the previous generation, with spatial multithreading( 88 cores / 176 threads), 1.2 TB / s LPDDR5X memory bandwidth with 3x more capacity and 1.8 TB / s NVLink-C2C bandwidth to GPUs( 2x previous).
• Third Generation Transformer Engine: Optimised acceleration for processing longcontext workloads, narrow-precision computations critical to scale modern AI workloads.
• Third Generation Confidential Computing: Delivers rack-scale confidential computing with a unified, GPU-level trusted execution environment that keeps models, data and prompts protected and isolated.
• Second Generation RAS Engine: Advanced reliability, availability and serviceability features, including real-time health checks without downtime. �
32 www. intelligentdatacentres. com