The Supermicro NVIDIA HGX platform powers many of the world's largest AI clusters, delivering computational output for today's transformative AI applications. Now featuring NVIDIA Blackwell Ultra, the 8U air-cooled system maximizes performance of eight 1100W HGX B300 GPUs with 2.3TB total HBM3e memory. Eight front OSFP ports with integrated ConnectX®-8 SuperNICs at 800 Gb/s enable turnkey deployment of NVIDIA Quantum-X800 InfiniBand or Spectrum-X™ Ethernet clusters. The 4U liquid-cooled system features DLC-2 technology with 98% heat capture, achieving 40% data center power savings. Supermicro Data Center Building Block Solutions® (DCBBS) and onsite deployment expertise provide complete solutions for liquid cooling, network topology and cabling, power delivery, and thermal management to accelerate AI factory time-to-online.

The Supermicro NVIDIA GB300 NVL72 tackles AI computational demands from training foundational models to large-scale reasoning inference. It combines high AI performance with Supermicro's direct liquid cooling technology, enabling maximum computing density and efficiency. Based on NVIDIA Blackwell Ultra, a single rack integrates 72 NVIDIA B300 GPUs with 288GB HBM3e memory each. With 1.8TB/s NVLink interconnects, the GB300 NVL72 operates as an exascale supercomputer in a single node. Upgraded networking doubles performance across compute fabric, supporting 800 Gb/s speeds. Supermicro's manufacturing capacity and end-to-end services accelerate liquid-cooled AI factory deployment and speed time-to-market for GB300 NVL72 clusters.

The new air-cooled NVIDIA HGX B200 8-GPU systems feature enhanced cooling architecture, high configurability for CPU, memory, storage, and networking, and improved serviceability from the front or rear side. Up to 4 of the new 8U/10U air-cooled systems can be installed and fully integrated in a rack, achieving the same density as the previous generation while providing up to 15x inference and 3x training performance. All Supermicro NVIDIA HGX B200 systems are equipped with a 1:1 GPU-to-NIC ratio supporting NVIDIA BlueField®-3 or NVIDIA ConnectX®-7 for scaling across a high-performance computing fabric.

The new front-I/O liquid-cooled 4U NVIDIA HGX B200 8-GPU system features Supermicro's DLC-2 technology. Direct liquid cooling now captures up to 92% of the heat generated by server components, such as CPU, GPU, PCIe switch, DIMM, VRM, and PSU, allowing up to 40% data center power savings and as low as a 50dB noise level. The new architecture further enhances efficiency and serviceability of the predecessor that was designed for NVIDIA HGX H100/H200 8-GPU systems. Available in 42U, 48U, or 52U configurations, the rack-scale design with the new vertical coolant distribution manifolds (CDM) means that horizontal manifolds no longer occupy valuable rack units. This enables 8 systems with 64 NVIDIA Blackwell GPUs in a 42U rack and up to 12 systems with 96 NVIDIA GPUs in a 52U rack.