Innovation & Tech

ASUS Ramps Vera Rubin NVL72 Output as Rivals Offer Same Rack

ASUS began volume shipments of NVIDIA Vera Rubin NVL72 racks in the fourth quarter and is pitching factory validation, digital-twin cooling modeling and zero-touch commissioning as its edge.

By Nathan Brooks

3 min read

ASUS Increases Vera Rubin NVL72 Output for Rack-Scale AI Deployments
ASUS Increases Vera Rubin NVL72 Output for Rack-Scale AI DeploymentsTumisu / pixabay

What's News

  • ASUS announced on November 17, 2026 that it is ramping fourth-quarter volume shipments of NVIDIA Vera Rubin NVL72 systems, citing robust market demand
  • The XA VR721-E3 AI POD is a fully liquid-cooled rack with 72 NVIDIA Rubin GPUs, 36 Vera CPUs and 227 kW peak power demand
  • A listed configuration includes nine NVLink 6 switches, 144 ConnectX-9 SuperNICs, 18 BlueField-4 DPUs and four 110 kW power shelves
  • ASUS detailed a six-layer validation framework on August 24, 2026, where any missed threshold blocks shipment until cause isolation, remediation and re-validation
  • The AI Factory workflow combines NVIDIA's DSX reference design with the Omniverse DSX Blueprint, introduced October 28, 2025, plus Schneider Electric, AVEVA and IBM

ASUS said on November 17, 2026 that it is ramping volume shipments of NVIDIA Vera Rubin NVL72 systems during the fourth quarter, citing robust market demand. The company is showing the platform at SC26 in Chicago inside a new AI Factory reference architecture spanning design simulation, factory validation, deployment automation and ongoing management.

With multiple system builders now offering Vera Rubin NVL72 platforms, the differentiation has moved off the hardware sheet and onto delivery. ASUS is betting that manufacturing consistency, pre-deployment testing and facility planning are what separate one supplier's racks from another's.

The display centerpiece is the XA VR721-E3 AI POD, a fully liquid-cooled rack holding 72 NVIDIA Rubin GPUs and 36 Vera CPUs, built for frontier-model training, complex reasoning and high-throughput inference. One listed configuration includes nine NVLink 6 switches, 144 ConnectX-9 SuperNICs, 18 BlueField-4 DPUs and four 110-kilowatt power shelves, with peak rack power demand of 227 kilowatts.

What happens before the rack ships?

At 227 kilowatts peak, the production process becomes part of the system's delivery risk. ASUS uses its Predictive Virtual Lab to assess potential signal-integrity problems during design, before volume manufacturing begins.

Automation then runs at the L10 node and L11 full-rack stages. Steps include optical inspection, protective-cover removal, module-positioning assistance and precision fastening, a set of steps ASUS says "reduces assembly variation."

The six-layer validation framework, detailed by ASUS on August 24, 2026, covers hardware functionality, firmware and remote-management checks, repeated power cycling and stress testing, NVIDIA diagnostic tools, and cross-GPU and network performance. Any test item that misses its threshold blocks the system from advancing while teams isolate the cause, remediate it and repeat validation. Powered aging assessments then track thermal behavior, power stability and system health before final inspection and shipment.

Why does the power and cooling work start in software?

ASUS's AI Factory workflow begins before a rack reaches the customer. It combines NVIDIA's DSX reference design with the Omniverse DSX Blueprint, which NVIDIA introduced on October 28, 2025.

Working with Schneider Electric, AVEVA and IBM, ASUS brings IT infrastructure, power, cooling and facility models into a simulation environment. Digital twins let designers co-design buildings, power, cooling and AI infrastructure, so constraints and design trade-offs surface before physical deployment rather than after.

In a May 13, 2026 technical account, ASUS described turning rack systems into validated simulation assets carrying power, thermal and connection-point metadata. Planners can then evaluate water temperature, flow and pressure, cooling-distribution-unit placement, piping and rack heat load alongside facility power constraints.

The point is practical: expose cooling or facility problems before equipment arrives, when changes can require expensive rework. A 227-kilowatt rack delivered into a building not designed for it is a facility project, not a server delivery.

What does this change for operators?

For buyers, the value is fewer surprises after the truck leaves. ASUS Professional Services covers site-readiness assessment, power and cooling planning, systems integration and on-site installation. The ASUS Infrastructure Deployment Center automates operating-system and software installation, cluster configuration and network setup. Its zero-touch workflow lets servers register, configure and join a cluster when powered on, with version-controlled scripts making deployments repeatable.

Once systems are running, ASUS Control Center Data Center Edition provides centralized infrastructure-health monitoring, power and cooling visibility, firmware and software updates, and security controls.

Alongside the NVL72 rack, ASUS is showing hybrid-cooled 3U and fully liquid-cooled 2U systems based on NVIDIA HGX Rubin NVL8, plus the Vera-CPU-based XA P2N-E2 for multi-step agent execution and orchestration. ASUS servers are available worldwide, though individual products and configurations vary by region.

The open question for the production expansion is whether ASUS can reproduce this chain of simulation, validation and automated commissioning consistently across customer sites. Its practical value lies in limiting costly facility changes, integration delays and GPU idle time as these densely integrated racks move into service, and the answer for operators is whether that consistency holds outside the SC26 demo floor in Chicago.

Source: unite.ai

Share this article:

More from Nathan Brooks

Nathan Brooks

Show full bio

News editor covering regulation, policy and their impact on entrepreneurship, startups and venture funding at Business Bearings.

2,251 articles

Related articles

« Previous articleNext article »