KR9288-E3 – 10U AI Training Server Supporting 8 GPUs or OAMs

The KR9288-E3 10U AI Training Server is built for next-generation AI workloads, including LLM training, HPC simulations, and generative AI applications.

Supporting 8 NVIDIA Blackwell Ultra GPUs (HGX™ B300) or Intel® Gaudi® 3 OAM accelerators, the system delivers maximum compute density and scalable performance. Equipped with 2x AMD EPYC™ 9005 processors, up to 6TB of DDR memory, high-speed CX8 IB interconnect, and hot-swap NVMe storage, it ensures high-throughput data processing and multi-node AI operations.

The KR9288-E3 is ideal for AI research centers, cloud AI deployments, and hyperscale HPC infrastructures, offering enterprise-grade reliability and flexible configuration options.

Main Application Areas:

• LLM Training: GPT-class and large-scale language models

• High-Performance Computing (HPC): Scientific simulations, modeling, and analytics

• Generative AI & Research: GPU/OAM-accelerated experiments and multi-node AI pipelines

• Data-Intensive Workloads: High-throughput AI inference and model fine-tuning

Price on request – daily updates apply. Please check availability before ordering.