AIXX Pedia › Model Comparison › H100 × H20
| Parameter | NVIDIA H100 · Hopper Generation | NVIDIA H20 · Hopper China-market Variant |
|---|---|---|
| Architecture | Hopper | Hopper (China-market Variant derived from the GH100) |
| Process | TSMC 4N | TSMC 4N (per media reports) |
| Board Memory | 80GB HBM3 (640GB across 8 cards) | 96GB HBM3 (768GB across 8 GPUs; upgraded version reportedly ~148GB HBM3e) |
| Memory Bandwidth | 3.35TB/s per card | 4.0TB/s per GPU (media reports, above the H100's 3.35TB/s) |
| GPU-to-GPU Interconnect | NVLink 4 · 900GB/s | NVLink 4 · 900GB/s (HGX module spec; PCIe standalone cards have no NVLink) |
| Board Power (TBP/TGP) | Up to 700W (configurable) | ~400W TDP (media reporting) |
| Form Factor & Cooling | HGX H100 8-card baseboard (mostly air-cooled towers · OEM liquid-cooled variants available) | HGX H20 8-GPU baseboard · PCIe standalone card (400W-class air-cooled tower) |
| Launch / Delivery | Announced 2022-03 · shipping since 2022-10 | Launched in 2024 (per media reports) |
| China-market Timeline | Subject to China export controls since 2022-10 · not on the approved list for China | China-market Variant · licensing required 2025-04 → approval to resume sales 2025-07 → supply fluctuating with policy shifts (per public reports) |
| Mainland China Street Price Band | Approx. 5.02M per system (2026-10-07 live market floor · 1 unit · whole-lot asking basis) | None currently in stock in this library (verified 2026-10-08); historical anchor: ~1.1M per 8-GPU system in 2025-02 (per Cailianshe reporting) |
Specs are based on NVIDIA official public documentation and the two entry pages; price bands reflect actual on-sale listings in this library's marketplace and fluctuate with inventory—see the entry page "price band" for the latest. A dash in both columns means the entry does not yet include this parameter.
AIXX Index (0–100): estimated by this library using a weighted public-spec model—memory 40% + interconnect 30% + ecosystem 30%, weights adjusted by scenario. 90+ flagship · 75+ Excellent · 60+ usable · below 60 not applicable. The method is transparent and verifiable and does not constitute a performance promise; figures match the published figures on the two entry pages, so this page does not score separately.
| Scenario Index | H100 | H20 |
|---|---|---|
| LLM Inference | 85 | 88 |
| Model Training / Fine-tuning | 88 | 55 |
| HPC / FP64 Computing | 92 | 40 |
| Cost-sensitive Inference | 90 | 85 |
| Large-memory scenarios (long context / full R1) | 58 | 90 |
| Condition / Reliability / Value Retention (Experience Index) | Pending hands-on testing | Pending hands-on testing |
"—" = the machine's entry page has no rating for this scenario; this page does not fill it in. The Experience Index (condition/reliability/value retention) is pending hands-on testing—inspection batch data is being accumulated and will light up unit by unit as Inspection Reports are published.
After this comparison, return to the entry pages to verify details and the latest price bands: