ASUS plans to increase shipments of NVIDIA Vera Rubin NVL72 systems in the fourth quarter of 2026, adding to the supply push for AI data centers while leaving the size of its increase undisclosed.
The company attributes the planned ramp to market demand but gave no shipment count or customer names. It is presenting the hardware and AI Factory services at SC26, a supercomputing conference in Chicago. The announcement follows EconoTimes’ earlier coverage of ASUS’ Q4 shipment plans.
NVIDIA’s fiscal third-quarter 2027 results say Vera Rubin NVL72 shipments began during the period ended Oct. 25, 2026. That marks the platform’s broader shipment status; NVIDIA’s results do not specify ASUS’ delivery volume.
At the show, ASUS is displaying its AI POD XA VR721-E3, a liquid-cooled rack that combines 72 NVIDIA Rubin GPUs with 36 NVIDIA Vera CPUs. ASUS markets the system for frontier-model training and for complex reasoning and high-throughput inference. NVIDIA’s specifications list the same GPU and CPU counts for its NVL72 platform.
ASUS says its factory uses automated L10/L11 production and engineering risk assessments before systems enter a six-part verification process. The checks cover node hardware and firmware as well as rack-level operating trials, diagnostics, performance reviews and powered-aging tests.
The services pitch extends from facility planning and digital-twin modeling, including work with Schneider Electric, AVEVA and IBM, to infrastructure provisioning, cluster setup and software installation. ASUS says its Infrastructure Deployment Center handles those deployment tasks. The company did not report measured savings or faster deployment times. It also describes delivery reliability, risk reduction and support as ways system builders can distinguish their NVL72 offerings; that is ASUS’ characterization of the market.
SC26’s exhibition at McCormick Place is scheduled to run through Thursday, Nov. 19. ASUS lists its booth as No. 2234.



