Supermicro HGX H200 Server 8U (AMD EPYC 9474F)

Fully configured 8U HGX H200 8-GPU server. Ships with two 48-core AMD EPYC 9474F processors, 2.3TB of DDR5 and 8x H200 141GB — ready for LLM training and inference. The Supermicro AS-8125GS-TNHR (Configuration OTO-2) is a high-performance 8U enterprise GPU server designed for artificial intelligence (AI), high-performance computing (HPC), generative AI, large language model (LLM) training, deep learning, and GPU-accelerated data center workloads. Equipped with 8× NVIDIA H200 Tensor Core GPUs, dual AMD EPYC 9474F processors, and 2,304GB DDR5 ECC system memory, this server delivers powerful parallel computing capabilities for demanding enterprise and research applications. Built on the NVIDIA HGX H200 architecture, the system features 1,128GB of HBM3e GPU memory, 4.8TB/s memory bandwidth per GPU, and high-speed NVIDIA NVLink and NVSwitch interconnect technology. Available through Billion Bison Technologies, this system is suitable for enterprises, AI infrastructure providers, GPU cloud operators, research institutions, and organizations deploying scalable high-performance computing environments. Prices are dynamic. Please confirm the latest price, configuration and availability with our team before ordering.