Nvidia 900-21010-0040-000 H200 NVL Graphics CardPassive Cooling 141 GB PCIe
Central Computer
Overview
2u rackmount server. Optimized for AI inference and enterprise compute workloads.
Key Features
- H200 with 141GB VRAM
- NVIDIA NVLink GPU interconnect
- 2u rackmount form factor
Ideal For
enterprise teams and developers and research institutions running large-scale AI training, multi-GPU workloads, and enterprise inference deployments.
FP64
34 TFLOPS
FP64 Tensor Core
67 TFLOPS
FP32
67 TFLOPS
TF32 Tensor Core2
989 TFLOPS
BFLOAT16 Tensor Core2
1,979 TFLOPS
FPl 6 Tensor Core2
1,979 TFLOPS
FPS Tensor Core2
3,958 TFLOPS
INTS Tensor Core2
3,958 TFLOPS
GPU Memory
141GB
GPU Memory Bandwidth
4.8TB/s
Decoders
7 NVDEC7 JPEG
Confidential Computing
Supported
Max Thermal Design Power (TDP)
Up to 600W (configurable)
Multi-Instance GPUs
Up to 7 MIGs @l 8GB each
Form Facto
PCIe
Interconnect
2- or 4-way NVIDIA NVLink bridge: 900GB/s PCle Gens: l 28GB/s
Server Options
NVIDIA MGX"' H200 NVL partner and NVIDIA-Certified Systems with up to 8 GPUs
NVIDIA Al Enterprise
Included
$29,999.99
Prices may vary. Verify on vendor site.
Quick Specs
- FP64
- 34 TFLOPS
- FP64 Tensor Core
- 67 TFLOPS
- FP32
- 67 TFLOPS
- TF32 Tensor Core2
- 989 TFLOPS
- BFLOAT16 Tensor Core2
- 1,979 TFLOPS
- FPl 6 Tensor Core2
- 1,979 TFLOPS
Tags
llm-traininggenerative-ai