Gigabyte NVIDIA H200 NVL 141GB HBM3e PCIe GPU Card

Brand: Gigabyte | Category: GPUs

SKU: GV-NH200NVL-141G | Part #: GV-NH200NVL-141G | MPN: GV-NH200NVL-141G

Contact for Pricing — Request a Quote

PCIe Gen5 x16 interface with server-grade 10-pin PCIe power connector determines chassis and PSU compatibility. The Gigabyte NVIDIA H200 NVL 141GB HBM3e PCIe GPU Card (part number GV-NH200NVL-141G) brings enterprise-class AI acceleration to demanding datacenter workloads. Built on NVIDIA's Hopper architecture (GH200), this full-height, full-length dual-slot accelerator delivers 141 GB of HBM3e memory paired with a 4480-bit memory bus and 4.8 TB/s bandwidth—enabling rapid processing of massive datasets. The card houses 16,896 CUDA cores across 132 streaming multiprocessors, supported by 4th generation Tensor Cores with multi-precision compute capabilities: FP8 reaches 3958 TOPS, FP16 and BF16 each deliver 1979 TFLOPS, TF32 provides 989 TFLOPS, and FP64 achieves 67 TFLOPS. Support for FP64, FP32, TF32, BF16, FP16, FP8, INT8, and INT4 precisions ensures flexibility across machine learning, scientific computing, and high-performance analytics applications.

AI infrastructure teams, machine learning engineers, and GPU-accelerated research institutions purchase the Gigabyte H200 NVL to overcome memory and bandwidth bottlenecks in transformer model training, large language model inference, and enterprise AI pipelines. The 2nd generation Transformer Engine accelerates sparsity and mixed-precision workloads critical to modern deep learning. Passive thermal design requires forced-air datacenter cooling infrastructure, positioning this card for purpose-built server environments rather than edge deployments.

Organizations evaluating this accelerator for production AI clusters should contact Omnixon Global to request pricing, availability, and configuration guidance for the GV-NH200NVL-141G. Our enterprise solutions team stands ready to support your RFQ and technical validation process.