NVIDIA H100 SXM
The Industry Standard for Enterprise AI Training
| Architecture | Hopper (GH100) |
|---|---|
| VRAM | 80GB HBM3 |
| Mem Bandwidth | 3.35 TB/s |
| FP8 Tensor | 3,958 TFLOPS |
| Form Factor | SXM5 |
| Tenancy | Dedicated |
| Location | Missoula, Montana |
Ideal workloads: Fine-tuning foundation models, training custom neural networks, high-throughput batch inference, computer vision training, scientific simulation.