Prnewswireยท2025-09-11 20:05Core Insights - Supermicro has announced the broad availability of its NVIDIA Blackwell Ultra solutions, including Plug-and-Play (PnP)-ready NVIDIA HGX B300 systems and GB300 NVL72 racks, aimed at enhancing AI infrastructure deployment globally [1][2][5] Product Offerings - The NVIDIA Blackwell Ultra systems are designed for high performance and compute density, featuring advanced cooling designs and the ability to utilize up to 1400W per GPU, achieving 50% greater inferencing performance and 50% more HBM3e capacity compared to previous models [2][3] - Supermicro's complete Blackwell Ultra portfolio includes over 10 SKUs, enabling diverse AI factory environments at any scale [5][6] Technological Advancements - The solutions leverage direct liquid cooling technology, advanced air cooling, and optimized I/O design, which contribute to doubling compute network bandwidth and enhancing AI factory productivity [3][4] - The GB300 NVL72 rack-scale system achieves 1.1 exaFLOPS dense FP4 compute performance, while the NVIDIA HGX B300 systems provide up to 7.5x performance gains over NVIDIA Hopper-based systems [4][6] Deployment and Integration - Supermicro offers turn-key delivery of pre-validated solutions at system, rack, and data center scale, addressing challenges in AI infrastructure such as network topology, power delivery, and thermal management [2][6] - The integration of hardware with infrastructure and application software, including NVIDIA AI Enterprise, optimizes AI performance and supports rapid deployment [6] Environmental Impact - Supermicro's Data Center Building Block Solutions (DCBBS) help customers save up to 40% power, reduce data center footprint by 60%, and decrease water consumption by 40%, leading to a 20% lower total cost of ownership (TCO) [6]