CoreWeave completed the industry-first bring-up and validation of an NVIDIA Vera Rubin NVL72 rack on 2026-06-01 (Livingston NJ; Dell PowerEdge XE9812-integrated): 72 Rubin GPUs, 36 Vera CPUs, ~3.6 EFLOPS NVFP4 (sparse), 216 TB/s NVLink 6 at shipping spec
1st rack validated 2026-06-01observed
| 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 | NVLink 6 rack bandwidth is NVIDIA's shipping spec (216 TB/s = 72 x 3.0 TB/s; Vera Rubin NVL72 product page). CoreWeave's June 2026 announcement quoted NVIDIA's pre-shipment 260 TB/s. The 3.6 EFLOPS figure is NVIDIA's sparse NVFP4 inference rating; its dense NVFP4 rating is 2.52 EFLOPS per rack. |
| Caveat | June 1 records the first validated rack. CoreWeave stood up a multi-rack Vera Rubin NVL72 cluster in early September 2026 and announced limited availability on 2026-09-30, with Cognition the first customer running production workloads; that is not an unrestricted general-availability announcement. Dell says its PowerRack with XE9812 servers goes from delivery to production in under 6.5 hours (Dell blog, 2026-06-01; no methodology published). The ~3.6 EFLOPS figure is NVIDIA's sparse NVFP4 rating from its Vera Rubin NVL72 product page. |
| As of | 2026-10 |
| Source | CoreWeave (2026-06-01 bring-up announcement; 2026-09-30 Cognition limited-availability post) |
| Review | checking…review by 2027-01-31 · fast cadence |
| Recorded changes | last 2026-10-15 · 3 revisions tracked |
| Claim id | coreweave-first-vera-rubin-nvl72-bringup |
Where the guide uses it
- 7.2 NVIDIA Accelerators: Hopper → Blackwell → Vera Rubin → Rubin Ultra → Feynman
- 7.14 Server & System Integration
← Full numbers register — every date-stamped figure in the guide, with revision history.