(858) 613-1818Quote list 0

NVIDIA Vera Rubin NVL72

NVIDIA's Rubin generation rack with 72 Rubin GPUs, 36 Vera CPUs, 20.7 TB of HBM4 at 1,400 TB/s and 3,600 PFLOPS sparse NVFP4. Register interest now.

Family
Rubin
Edition
Rack scale system, 72 Rubin GPUs and 36 Vera CPUs
Memory
288 GB HBM4 per GPU, 20.7 TB HBM4 per rack, up to 54 TB LPDDR5X CPU memory
Memory bandwidth
19.2 TB/s per GPU, 1,400 TB/s per rack as published for the NVL72
Interconnect
Sixth generation NVLink, 3 TB/s per GPU, 216 TB/s per rack, 65 TB/s NVLink-C2C
Form factor
Single rack, 72 GPU NVLink domain with ConnectX-9 SuperNICs and BlueField-4 DPUs
TDP
Not yet published
Cooling
Liquid
Best for
Next generation AI factories planning Rubin deployments from 2026 onward
Availability
Coming soon

Vera Rubin NVL72 is the successor to GB300 NVL72. NVIDIA's published figures give 72 Rubin GPUs with 288 GB of HBM4 each for 20.7 TB per rack at 1,400 TB/s, 36 Vera CPUs with 3,168 custom Olympus cores and up to 54 TB of LPDDR5X, sixth generation NVLink at 216 TB/s, 65 TB/s of NVLink-C2C, 32.4 TB/s of bi-directional networking, and 3,600 PFLOPS of sparse NVFP4 inference or 2,520 PFLOPS of dense NVFP4 training.

NVIDIA states that Vera Rubin is ramping into full production with its system partners. Per GPU power, rack power and delivery schedules for reseller channels are not yet published, so we are taking expressions of interest rather than firm orders. If you are planning a Rubin deployment, we can reserve a place in the queue, map the site requirements against your data center, and quote an interim GB300 NVL72 or HGX B300 cluster to bridge the gap.

Systems that carry this GPU

Want this GPU in a different chassis?

We quote every class that carries it, from a single node to a pod. Tell us the workload and we propose the chassis.

Request a quotationView quote list