Product
NVIDIA H200 NVL 141GB
Part number NVIDIA H200 NVL
- Price
- USD $100
- Stock
- 1
Hopper PCIe accelerator with 141 GB of HBM3e memory at 4.8 TB/s, a dual-slot air-cooled form factor, and a configurable TDP up to 600 W, for MGX and NVIDIA-Certified servers.
- 141 GB HBM3e memory at 4.8 TB/s
- PCIe dual-slot, air-cooled card
- Configurable TDP up to 600 W
- 2- or 4-way NVLink Bridge at 900 GB/s per GPU
- NVIDIA AI Enterprise included, per NVIDIA’s published H200 table
Specifications
| GPU Model | NVIDIA H200 NVL |
|---|---|
| Architecture | NVIDIA Hopper |
| Form Factor | PCIe dual-slot air-cooled |
| GPU Count | 1 |
| GPU Memory | 141 GB HBM3e |
| Memory Bandwidth | 4.8 TB/s |
| Interconnect | 2- or 4-way NVLink Bridge 900 GB/s per GPU; PCIe Gen5 128 GB/s |
| Max TDP | Up to 600 W (configurable) |
| Cooling | Passive, dual-slot air-cooled |
| FP64 | 30 TFLOPS |
| FP64 Tensor Core | 60 TFLOPS |
| FP32 | 60 TFLOPS |
| TF32 Tensor Core | 835 TFLOPS (with sparsity) |
| FP16 / BFLOAT16 Tensor Core | 1,671 TFLOPS (with sparsity) |
| FP8 Tensor Core | 3,341 TFLOPS (with sparsity) |
| INT8 Tensor Core | 3,341 TOPS (with sparsity) |
| Multi-Instance GPU | Up to 7 MIGs at 16.5 GB each |
| Confidential Computing | Supported |
| Typical Host | NVIDIA MGX H200 NVL partner and NVIDIA-Certified Systems, up to 8 GPUs |
Description
The NVIDIA H200 NVL puts the H200 memory subsystem into a dual-slot, air-cooled PCIe card. It keeps the 141 GB of HBM3e and 4.8 TB/s of memory bandwidth published for H200 SXM, but is intended for lower-power, air-cooled enterprise racks that need flexible GPU counts rather than a fixed HGX SXM baseboard.
NVIDIA lists a configurable thermal design power of up to 600 W and a PCIe Gen5 host interface at 128 GB/s. Two- or four-way NVLink Bridges provide 900 GB/s per GPU, so a small group of cards can share memory traffic without an NVSwitch board. Peak Tensor rates published for the NVL variant are slightly below the SXM figures (for example 1,671 TFLOPS of FP16/BF16 Tensor performance with sparsity, versus 1,979 TFLOPS on H200 SXM). MIG is available with up to seven instances at 16.5 GB each. NVIDIA’s product table states that NVIDIA AI Enterprise is included with H200 NVL.
This card is a strong fit for inference and mixed enterprise workloads that need H200-class memory in a conventional server, including MGX H200 NVL partner systems. For the highest-density, NVSwitch-connected topology, use the H200 SXM module in an HGX H200 host such as the GIGABYTE G593-SD1. Request a quote for the NVL card or a matching air-cooled server.
Related products
NVIDIA H200 SXM 141GB
P/N NVIDIA H200 SXM
Price USD $100 Stock 1
Hopper accelerator with 141 GB of HBM3e memory at 4.8 TB/s, an SXM form factor, and a configurable TDP up to 700 W, designed for HGX H200 training and inference nodes.
- GPU Model
- NVIDIA H200 SXM
- Form Factor
- SXM
- GPU Count
- 1
- GPU Memory
- 141 GB HBM3e
NVIDIA H100 80GB PCIe
P/N NVIDIA H100 80GB PCIe
Price USD $100 Stock 1
Hopper-architecture PCIe Gen5 accelerator with 80 GB of HBM2e memory and a 350 W envelope, suited to mainstream servers and one- or two-GPU inference nodes.
- GPU Model
- NVIDIA H100 80GB PCIe
- Form Factor
- PCIe Gen5 dual-slot
- GPU Count
- 1
- GPU Memory
- 80 GB HBM2e
Ready to specify a configuration?
Share the use case, the target build, quantity, destination country, and timeline. We will take it from the inquiry.