
NVIDIA H100 80GB PCIe
The workhorse of production inference: H100 PCIe drops into standard 2U/4U chassis without exotic cooling and runs the full CUDA / TensorRT-LLM stack. Allocation batches land in Singapore regularly - reserve against PO.
MOQ 1 unit
| Architecture | NVIDIA Hopper (GH100) |
| CUDA cores | 14,592 |
| Tensor cores | 456 (4th gen) |
| Memory | 80 GB HBM2e |
| Memory bandwidth | 2.0 TB/s |
| FP8 performance | 3,026 TFLOPS (with sparsity) |
| FP16 performance | 1,513 TFLOPS (with sparsity) |
| FP64 performance | 26 TFLOPS |
| Interconnect | NVLink 600 GB/s + PCIe 5.0 x16 |
| MIG support | Up to 7 instances |
| TDP | 350 W |
| Form factor | Dual-slot FHFL, passive cooling |
| Warranty | 12 months, SYNDROME AG |
The workhorse of production inference: H100 PCIe drops into standard 2U/4U chassis without exotic cooling and runs the full CUDA / TensorRT-LLM stack. Allocation batches land in Singapore regularly - reserve against PO.
Current inventory: 6 units across two free-port hubs. Pricing is FOB Singapore / Dubai in USD; volume orders from 10 units receive −5…10%. Serial numbers are recorded for every unit before it leaves the warehouse.
Air and sea freight from Singapore or Dubai under EXW, FOB, CIF or DDP (Incoterms 2020). Export clearance and shipping documents are handled by our hubs. Commercial terms are structured individually for each deal and jurisdiction — your manager will walk you through the options.
New hardware carries a 12-month warranty (refurbished — 6 months). RMA is handled through the nearest hub with advance replacement from stock where available. Serial numbers are recorded for every unit that leaves the warehouse.
