The KR9288 is a 10U high-performance AI computing platform engineered for Large Language Model (LLM) training and multimodal inference workloads. Powered by the latest NVIDIA Blackwell HGX™ B300 8-GPU NVLink module, it delivers unmatched compute density and bandwidth for advanced AI models. It also supports OAM accelerators such as Intel® Gaudi®.
With support for Intel® Xeon® 6 or AMD EPYC™ 9005 processors, flexible memory configurations, and robust network expansion, the KR9288 is designed to handle the most demanding AI deployments. An advanced power and thermal design ensures stability under maximum load.
Highlights
✔ Extreme GPU Power – 8x NVIDIA B300 GPUs with 2.3TB HBM3e and 64TB/s bandwidth
✔ Up to 50% higher FP4 performance than the previous generation
✔ Redundant Power Delivery – 12V/54V separation, N+1/N+N PSU configurations
✔ High Network Density – 8x CX8 InfiniBand + 4x full-height N-S fabric
✔ Flexible Storage Options – up to 8x NVMe U.2 + 2x NVMe/SATA M.2 + optional RAID
✔ Efficient Cooling – Optimized airflow with front I/O design for improved reliability
Price on request – subject to daily changes. Please check availability before ordering.
