Super Micro Computer introduces liquid-cooled NVIDIA HGX B300 systems

Super Micro Computer, Inc. introduced liquid-cooled NVIDIA HGX B300 systems and said they were intended for high-density hyperscale and Artificial Intelligence (AI) factory deployments because of their Graphics Processing Unit (GPU) density and power efficiency.

Supermicro positioned the new systems within its Data Center Building Block Solutions, noting they addressed Total Cost of Ownership (TCO), serviceability, and operational efficiency and that validated rack configurations aimed to shorten time-to-online for large deployments.

The company described two hardware formats and their cooling and density characteristics: a 4U front I/O system for 19-inch Environmental Impact Assessment (EIA) racks and a compact 2-OU (OCP) system built to the 21-inch OCP Open Rack V3 specification. The 4U option supported up to 64 GPUs per EIA rack with DLC-2 liquid cooling that captured up to 98% of system heat, while the 2-OU design supported up to 144 GPUs per ORV3 rack with blind-mate manifold connections, modular GPU/CPU trays, and eight NVIDIA Blackwell Ultra GPUs at up to 1,100W Thermal Design Power (TDP) each.

Supermicro outlined rack-scale scaling and networking details, including compatibility with NVIDIA Quantum-X800 InfiniBand switches, Supermicro 1.8MW in-row coolant distribution units, and a SuperCluster example combining eight compute racks, three networking racks, and two in-row CDUs for 1,152 GPUs. The company also cited 2.1TB of HBM3e GPU memory per system and doubled fabric throughput up to 800Gb/s via integrated NVIDIA ConnectX-8 SuperNICs when used with NVIDIA Quantum-X800 InfiniBand or NVIDIA Spectrum-4 Ethernet.

“With AI infrastructure demand accelerating globally, our new liquid-cooled NVIDIA HGX B300 systems deliver the performance density and energy efficiency that hyperscalers and AI factories need today,” said Charles Liang, president and CEO of Supermicro. “We're now offering the industry's most compact NVIDIA HGX B300 solutions—achieving up to 144 GPUs in a single rack—while reducing power consumption and cooling costs through our proven direct liquid-cooling technology. Through our DCBBS, this is how Supermicro enables our customers to deploy AI at scale: faster time-to-market, maximum performance per watt, and end-to-end integration from design to deployment.”

The company said the systems were available for high-volume shipment, that Data Center Building Block Solutions were delivered as fully validated L11 and L12 rack solutions, and that the new products expanded Supermicro's NVIDIA Blackwell portfolio to include the NVIDIA GB300 NVL72, NVIDIA HGX B200, and NVIDIA RTX PRO 6000 Blackwell Server Edition.