NVIDIA GPU Accelerator — H200 NVL PCIe
- Part Number
- H200 NVL PCIe
- Manufacturer
- NVIDIA
- Condition
- Any
- Category
- Spare Parts & Accessories
In stock
The NVIDIA H200 NVL PCIe is a Hopper-architecture data center GPU accelerator that pairs 141 GB of fast HBM3e memory with 4.8 TB/s of memory bandwidth on a dual-slot, air-cooled PCIe Gen5 card with a 600 W max power envelope. Built around 16,896 CUDA cores and 528 Tensor Cores, and scalable in 2- or 4-way clusters via NVLink bridges at 900 GB/s per GPU, it targets large AI model training, LLM inference and HPC workloads in servers with up to eight GPUs. We supply the H200 NVL to integrators, resellers and enterprises expanding GPU compute capacity. Contact us for current availability, condition and pricing.
Specifications
| Architecture | NVIDIA Hopper |
|---|---|
| CUDA Cores | 16,896 |
| Form Factor | Dual-slot PCIe, air-cooled |
| GPU Memory | 141 GB HBM3e |
| Interface | PCIe Gen5 x16 |
| Interconnect | 2- or 4-way NVLink bridge, 900 GB/s per GPU |
| Max Power (TDP) | Up to 600 W (configurable) |
| Memory Bandwidth | 4.8 TB/s |
| Multi-Instance GPU (MIG) | Up to 7 instances @ 16.5 GB each |
| Part Type | GPU Accelerator |
| Tensor Cores | 528 |
| Brand | NVIDIA |
| Part Number | H200 NVL PCIe |
| Condition | Any |
| Availability | In stock — bulk pricing on request |
Similar Equipment
More NVIDIA H200 equipment we stock