Pictures may differ from actual products
PNY NVIDIA H200 NVL
141 GB HBM3e - PCI Express 5.0 x16 (NVIDIA H200 NVL)
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
GPU dedicated to generative AI and high-performance computing
The PNY NVIDIA H200 NVL professional graphics card sets a new standard for professional GPUs dedicated to generative artificial intelligence and high-performance computing. Designed for enterprise data centre environments, this card leverages the NVIDIA Hopper architecture, significantly boosting performance for large language model (LLM) inference, advanced AI system acceleration and the most demanding HPC applications. With massive memory, record-breaking bandwidth and optimised software compatibility, it is an essential tool for organisations looking to deploy AI at scale.
High processing speed with HBM3e memory
At the heart of the innovation, the H200 NVL features high-capacity HBM3e memory that delivers unmatched processing speed and efficiency for AI workloads. This larger, faster memory accelerates next-generation language models and inference tasks up to 1.7× faster than the previous generation, while reducing operational costs through improved energy efficiency. It is ideal for intensive AI deployments, from text and multimedia content generation to data analysis and intelligent automation.
High-performance design
Designed for enterprise servers with space constraints and air-cooled architectures, the PNY NVIDIA H200 NVL offers configuration flexibility ideal for standard racks. It comes with a five-year subscription to NVIDIA AI Enterprise, providing a comprehensive software suite for developing, integrating and deploying production-ready AI applications. This combination of hardware and software power reduces the time to market for AI solutions while ensuring stability, security and scalability.
Key features:
- NVIDIA Hopper-based GPU architecture, designed for generative AI and HPC
- GPU memory: 141 GB HBM3e with ultra-high bandwidth
- Memory bandwidth: 4.8 TB/s
- System interface: PCI Express 5.0 x16
- GPU interconnect: NVIDIA NVLink for multi-GPU configurations (x2 or x4)
- Software support: 5-year NVIDIA AI Enterprise subscription included
- Form factor designed for enterprise racks with air cooling
- Designed to accelerate LLM inference, machine learning and HPC
- FP64: 34 TFLOPS, FP64 Tensor Core: 67 TFLOPS, FP32: 67 TFLOPS, FP32 Tensor Core: 989 TFLOPS, BFLOAT16 Tensor Core: 1,979 TFLOPS, FP16 Tensor Core: 1.979 TFLOPS, FP8 Tensor Core: 3.958 TFLOPS, INT8 Tensor Core: 3.958 TFLOPS
Select your criteria
General information |
Product name |
PNY NVIDIA H200 NVL |
Brand |
PNY | |
Model |
TCSH200NVLPCIE-PB | |
Graphics chipset |
Graphics chipset |
NVIDIA H200 NVL |
Number of GPU |
1 | |
Bus |
PCI Express 5.0 x16 | |
Memory |
Video memory size |
141 GB |
Memory interface |
6016 bit(s) | |
Video memory frequency |
3201 MHz | |
Memory type |
HBM3e | |
Applications |
Multi-GPU |
None |
Use |
Computing station | |
Display |
Dual monitor |
No |
HD Ready |
No | |
TV tuner |
No | |
Connectivity |
Video outputs |
None |
Design |
Cooling type |
Passive (fanless) |
Consumption |
600 W | |
Low profile |
No | |
Format |
Dual Slot | |
Power connector(s) |
1 X PCI Express 16 Pin (12VHPWR) | |
Length |
254 mm | |
Warranties |
Commercial warranty |
Seller 2 years
|
Legal warranty |
See terms & conditions
|
|
Safety Information |
Responsible person |
PNY Technologies Europe |
Postal address |
ZAC des Belles Vues 1 Rue de la Fosse aux Loups 91140 Villejust France | |
Electronic Address |
Click here
|
See all PNY products
Product referenced on 27 Feb 2026
For the moment, there are no customer reviews.Write a review
In the same universe: h200