NVIDIA Vera Rubin NVL72 is a rack-scale AI-computing system built on the third-generation NVIDIA MGX NVL72 rack design, integrating six new chips for agentic AI workloads. The rack connects 72 Rubin GPUs and 36 Vera CPUs across 18 compute trays and 9 NVLink switch trays, alongside ConnectX-9 SuperNICs, BlueField-4 DPUs and Spectrum-6 Ethernet switches. Each Rubin GPU includes up to 288 GB of HBM4, and sixth-generation NVLink delivers 3.6 TB/s per GPU and 260 TB/s of scale-up bandwidth per rack. A PCB midplane replaces the cabling, hoses and fans of prior compute trays, and the system is fully liquid-cooled at 45°C inlet. Dynamic Max-Q power provisioning and Intelligent Power Smoothing, with 400 J of per-GPU energy storage, reclaim stranded power. NVIDIA Confidential Computing spans the rack, and BlueField-4 DPUs add zero-trust security. Vera Rubin NVL72 reaches general availability in the second half of 2026.