SuperX has launched its latest breakthrough in AI infrastructure — the GB300 NVL72 system, a next-generation, liquid-cooled AI supercomputing platform designed to meet the immense performance and efficiency demands of modern data centers. The announcement marks a significant leap in scalable AI hardware as enterprises and research institutions race to power increasingly complex AI models.
Powered by NVIDIA GB300 Grace Blackwell Chips
The GB300 NVL72 integrates NVIDIA’s cutting-edge GB300 Grace Blackwell chips, offering unprecedented computing performance for large-scale AI training, inference, and analytics workloads. Built for hyperscale environments and enterprise deployment, the system delivers massive parallel processing capabilities while maintaining exceptional energy efficiency through its advanced liquid-cooling design.
“Our GB300 NVL72 platform is built for the next era of AI computing,” said a SuperX spokesperson during the product reveal. “By leveraging NVIDIA’s most powerful chips and our custom cooling and integration design, we’re enabling customers to push the limits of AI performance without compromising on sustainability.”
Integrated Design for Efficiency and Scale
The GB300 NVL72 combines compute, networking, and cooling into a single, optimized rack-scale system. This unified approach simplifies deployment, minimizes latency, and enhances operational efficiency for organizations scaling AI workloads across vast datasets. The system supports high-bandwidth interconnects and a unified memory architecture, enabling seamless collaboration between AI models and data pipelines.
Built for Multi-Trillion Parameter AI Models
Early demonstrations of the GB300 NVL72 showcased its ability to train and deploy multi-trillion parameter AI models with remarkable speed and reliability. The platform’s high-density architecture and advanced cooling solution ensure consistent performance under heavy computational loads, positioning it as one of the most capable AI systems in the market.
Driving the Future of AI Infrastructure
SuperX’s unveiling of the GB300 NVL72 underscores the rapid evolution of AI hardware, as global tech firms race to build infrastructure capable of supporting next-generation machine learning and generative AI applications. The system’s rack-scale integration and focus on energy efficiency reflect the growing trend toward sustainable, high-performance computing.
The GB300 NVL72 is expected to begin shipping to select data center and research customers in early 2026, with broader availability anticipated later in the year.
