Suppliers with verified business licenses
Audited Supplier Audited by an independent third-party inspection agency
The NVIDIA H100 96GB PCIe GPU is the most advanced data center GPU designed for AI training, inference, and high-performance computing (HPC). Built on the NVIDIA Hopper architecture, the H100 PCIe delivers unprecedented performance with 4th generation Tensor Cores, Transformer Engine, and NVLink support.
With 96GB of high-bandwidth HBM3 memory, the H100 PCIe provides massive throughput and efficiency for the most demanding AI workloads, including large language models, scientific simulations, and enterprise data analytics. It is the ideal choice for hyperscale data centers and research institutions requiring next-generation acceleration.
| Item | Specification |
|---|---|
| GPU Architecture | NVIDIA Hopper |
| Form Factor | PCIe Gen5 |
| CUDA Cores | 14,592 |
| Tensor Cores | 4th Generation |
| Transformer Engine | Yes |
| Memory | 96GB HBM3 |
| Memory Bandwidth | 3.0 TB/s |
| NVLink | 4 Links, 600 GB/s total |
| Interface | PCI Express Gen5 x16 |
| Max Power Consumption | 350W |
| Cooling | Passive cooling (server integration required) |
| Form Factor | Dual-slot PCIe |
| Multi-GPU Scalability | Supported with NVLink and NVSwitch |
| ECC Support | Yes |
AI training for large language models (LLMs)
Deep learning and machine learning workloads
High-performance computing (HPC)
Scientific simulations and big data analytics
Cloud computing and enterprise AI solutions
Data center GPU acceleration
Standard anti-static and protective packaging
Worldwide shipping via DHL / FedEx / UPS / EMS / Air / Sea
Delivery time: 3-7 working days for stock
Condition: 100% Original / Brand New
Warranty: 1 Year (Extended warranty optional)
MOQ: 1 Piece
Model Number: NVIDIA H100 96GB PCIe
){kind=link}