SemiAnalysis Oct 2024 1,024-H100 model: back-end cost saving from combined virtual modular switch placement and 2:1 oversubscription versus NVIDIA reference architecture
31.6% combined VMS + 2:1 modeled savingmodeled
| Value kind | modeled — Interpret this value according to its displayed kind, scope, source, and as-of date. |
|---|---|
| Scope | SemiAnalysis model for 1,024 H100s: combined virtual modular switch placement and 2:1 oversubscription versus NVIDIA reference architecture. The non-blocking virtual modular switch alone saved 24.9%. |
| As of | 2024-10 |
| Source | SemiAnalysis AI Neocloud Playbook |
| Review | checking…review by 2027-04-01 · fast cadence |
| Recorded changes | last 2026-10-05 · 3 revisions tracked |
| Claim id | training-back-end-fabric-non-blocking-2-1 |
Where the guide uses it
← Full numbers register — every date-stamped figure in the guide, with revision history.