
ASUS has begun ramping up volume shipments of systems built on NVIDIA’s Vera Rubin platform, as demand for AI infrastructure continues to surge.
The company said on Tuesday it is increasing production of its NVIDIA Vera Rubin NVL72 systems in the fourth quarter, and is showing off its AI Factory reference architecture at SC26, the supercomputing conference in Chicago.
A full rack of AI power
At the centre of the showcase is the ASUS AI POD XA VR721-E3, a rack-scale system that packs 72 NVIDIA Rubin graphics processors and 36 NVIDIA Vera central processors into a single fully liquid-cooled rack. It is designed for training frontier AI models, complex reasoning and high-throughput inference — the heaviest workloads in modern AI.
Vera Rubin is NVIDIA’s latest AI platform, succeeding the Blackwell generation. The NVL72 designation refers to a rack containing 72 graphics processors working as a single system, a design aimed at data centres building large-scale AI capacity.
From factory floor to data centre
ASUS says its advantage is not just the hardware. The company offers end-to-end AI Factory services, covering digital-twin planning, deployment automation and lifecycle operations, with L10 and L11 manufacturing — the final stages of server assembly and testing — to make rack-scale deployments more predictable.
The company is presenting the technology at Booth 2234 at SC26, which runs from November 17 to 19 at McCormick Place in Chicago.
Why it matters
Demand for AI computing has outstripped supply for much of the past two years, as companies race to train larger models and run AI agents at scale. NVIDIA’s platforms sit at the centre of that build-out, and system makers such as ASUS turn the chips into deployable data-centre infrastructure.
Ramping shipments in the fourth quarter signals that the Vera Rubin generation is moving from launch into volume production — and that data-centre operators are buying it in quantity.
Sources: ASUS / PR Newswire.



