NVLink5 scale-up (1.8 TB/s/GPU bidirectional, ~900 GB/s/dir; 130 TB/s NVL72 rack) over ~400G scale-out NIC (~50 GB/s/dir) — keep TP/EP inside scale-up
~18×observed
| Value kind | observed — Reported measurements, counts, and specifications keep the precision and scope stated by their source; an exact specification is not treated as a range. |
|---|---|
| Scope | The ~18x ratio pairs GB200 (NVLink 5, ~900 GB/s per direction) with a 400G NIC (~50 GB/s per direction). Each generation pairs with its own NIC: GB300 is ~9x (NVLink 5 against an 800G ConnectX-8), and Vera Rubin is ~7.5x (NVLink 6 at the 3 TB/s shipping spec, ~1.5 TB/s per direction, against a 1.6T ConnectX-9 at ~200 GB/s per direction). Across generations the gap is about an order of magnitude, 8-18x. |
| As of | 2025 |
| Source | NVIDIA NVLink; SemiAnalysis |
| Review | checking…review by 2026-11-23 · standard cadence |
| Recorded changes | last 2026-07-02 |
| Claim id | nvlink5-scale-up-1-8-tb-s-gpu-bidirectional-900 |
Where the guide uses it
← Full numbers register — every date-stamped figure in the guide, with revision history.