SuperX Launches Rack-Scale AI Platform Powered by NVIDIA GB300 Chips, Redefining Data Center Infrastructure

Core Insights - SuperX AI Technology Limited has launched the SuperX GB300 NVL72 System, a cutting-edge AI supercomputing platform utilizing the NVIDIA GB300 Grace Blackwell Ultra Superchip, aimed at overcoming the limitations of traditional data center infrastructure [1][2][3] Product Overview - The SuperX GB300 NVL72 System delivers up to 1.8 exaFLOPS of AI performance in a single liquid-cooled rack, representing a significant advancement in compute density and energy efficiency [2][5] - The system integrates 72 NVIDIA Blackwell Ultra GPUs and 36 NVIDIA Grace CPUs, achieving a chip-to-chip link bandwidth of 900GB/s and a total memory of 165TB HBM3E and 17TB LPDDR5X [4][6] Technological Advancements - The system's design necessitates advanced power solutions, particularly 800 Voltage Direct Current (800VDC), to ensure stability and operational viability for next-generation workloads [3] - The GB300 System is optimized for massive horizontal scaling, allowing for the connection of multiple GPUs into a single system, thus redefining standards for large-scale AI training and inference [5][6] Market Positioning - SuperX positions the GB300 NVL72 System as a foundational infrastructure for organizations developing future AI capabilities, catering to sectors such as hyperscale computing, scientific research, and industrial applications [7][10]