Specifications

Overview

The NVIDIA H100 NVL 94GB (900-21010-0020-000) pairs the Hopper GPU with 94 GB of HBM3 and 3.9 TB/s of memory bandwidth — the highest-capacity single-GPU H100 configuration. It targets large-language-model inference, retrieval-augmented generation, and multi-tenant GPU serving where model weight capacity is the binding constraint. QS Compute supplies H100 NVL 94GB cards in bulk for GPU server builders.