The Definitive Guide toAI Data Centers
Ask the GuideAboutAccount
GuideGlossaryNVLink

NVLink

NVIDIA's high-bandwidth GPU-to-GPU interconnect for tying many GPUs into one memory-coherent scale-up domain.

Current numbers

3.6 TB/sNVLink 6 per-GPU bandwidth (Rubin); 900 GB/s Hopper, 1.8 TB/s Blackwell — ~260 TB/s per NVL72 rackas of H2 2026 (announced) · register ↗
1st rack validated 2026-06-01CoreWeave industry-first Vera Rubin NVL72 bring-up (Jun 1 2026, Livingston NJ; Dell-integrated) — 72 Rubin GPUs + 36 Vera CPUs, ~3.6 EFLOPS FP4, 260 TB/s NVLink 6; a single validated rack, not fleet GAas of 2026-06 · register ↗
1st rack validated 2026-06-01first Vera Rubin NVL72 rack validated Jun 2026 (CoreWeave/Dell, Livingston NJ) — ~3.6 EFLOPS FP4, 260 TB/s NVLink 6; a single rack, not fleet GAas of 2026-06 · register ↗

← All terms