ASUS ramps up NVIDIA Vera Rubin NVL72 shipments and brings AI Factory to SC26

ASUS announced on Monday (17) that it is ramping up volume shipments of NVIDIA Vera Rubin NVL72 systems in the fourth quarter of 2026, citing robust demand for the hardware. The company is also showcasing its AI Factory reference architecture at SC26, the supercomputing conference running November 17–19 at McCormick Place in Chicago, at booth 2234.

At the center of the showcase is the ASUS AI POD XA VR721-E3, a fully liquid-cooled rack-scale system integrating 72 NVIDIA Rubin GPUs and 36 NVIDIA Vera CPUs, aimed at frontier-model training, complex reasoning and high-throughput inference. The demonstration also includes trays with NVIDIA Groq 3 LPX and Vera CPUs, the NVIDIA Spectrum-X SN6600 Ethernet switch and NVIDIA CMX, combining compute, networking and context memory.

From factory to validated rack

According to ASUS, what sets NVL72 integrators apart is increasingly delivery reliability rather than access to silicon. The company points to its Predictive Virtual Lab, which flags signal-integrity risks early in the design phase, L10/L11 manufacturing automation — including automated optical inspection, protective-cover removal, module-positioning assistance and precision fastening — and a six-layer validation framework, from node level (L10) to the full rack (L11), covering hardware functionality, BIOS and BMC checks, power cycling, stress testing, NVIDIA diagnostics and rack-level performance evaluation, plus powered aging assessments.

Beyond the NVL72

ASUS is also presenting two NVIDIA HGX Rubin NVL8-based systems — the 3U hybrid-cooled XR R1I12B-M1BHFS and the 2U fully liquid-cooled XR R1I12B-M1BLFS. The NVL8 systems had already been announced by the company in October. The MGX portfolio also includes the XA P2N-E2, powered by NVIDIA Vera CPUs for multi-step agent execution and orchestration, and the ESC8000A-E13P, with NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs.

On the data side, the showcase includes the VS320D-RS26N enterprise storage platform, the high-density OJ340A-RS60 object storage system and the compact 1U RS500A-E14B-R12U server, powered by 6th Gen AMD EPYC 9006 processors with PCIe 6.0 connectivity and NVMe storage.

From design to operations

For AI factory planning, ASUS combines its AI Factory solution with the NVIDIA DSX reference design and the NVIDIA Omniverse DSX Blueprint for AI Factory Digital Twins, part of NVIDIA DSX Sim, to simulate facilities and configurations before physical deployment, working with partners including Schneider Electric, AVEVA and IBM. The company's professional services cover site assessment, power and cooling, installation and the ASUS Infrastructure Deployment Center (AIDC), which automates provisioning and cluster configuration, plus ASUS Control Center (ACC) Data Center Edition for centralized monitoring, security and update management.

Context

Geekiki has already covered the Vera Rubin NVL72 platform in detail — including its specifications and throughput-per-dollar gains — and its MLPerf debut, with up to 2x performance. Today's announcement marks the platform's move from specifications and benchmarks to volume delivery.

Image credit: ASUS, via PR Newswire — the ASUS AI Factory booth at SC26, featuring the NVIDIA Vera Rubin NVL72 display.

Sources

Transparency note: this article was produced with the assistance of artificial intelligence, cross-checked against public posts on X and sources from the internet. Please consult the original sources listed above to verify the details.

Por Muse