NVIDIA HGX A100 8-GPU AI/HPC Server (Used)

GIGABYTE G492-ZD0 – NVIDIA HGX A100 8-GPU AI/HPC Server (Used) 8× NVIDIA A100 80GB SXM4 | 640GB HBM2e GPU Memory | Dual AMD EPYC 7742 | 512GB DDR4 RAM | 4U Rackmount Product Overview The GIGABYTE G492-ZD0 is a high-performance 4U GPU server engineered for artificial intelligence, deep learning, high-performance computing (HPC), large language model (LLM) applications, and enterprise data center workloads. This used, fully configured system features 8× NVIDIA A100 80GB SXM4 Tensor Core GPUs, dual AMD EPYC 7742 processors, 512GB DDR4 memory, and eight NVIDIA ConnectX-7 200Gbps network adapters. Built on the NVIDIA HGX A100 architecture, the server provides 640GB of combined HBM2e GPU memory and high-bandwidth GPU interconnect technology, supporting demanding AI training, inference, and distributed computing applications. Powered by dual 64-core AMD EPYC 7742 processors, the system offers 128 CPU cores and 256 threads, providing substantial CPU computing capacity alongside GPU acceleration. Available through Billion Bison Technologies, this pre-owned GPU server is suitable for businesses seeking enterprise AI and HPC infrastructure without the cost of purchasing a comparable new system.