South Korea – ASUS has set out a new direction for its AI infrastructure business, positioning itself as a builder and operator of “AI factories” rather than a supplier of standalone systems, according to the company.
The announcement was made at ASUS AI Tech 2026 in Seoul, South Korea, where the company outlined its expanded collaboration with NVIDIA and a wider ecosystem of technology partners.
Among the partners named are Aleria, AMD, Crusoe, Foxlink, IBM, Intel, Samsung, Schneider Electric, and WD, ASUS said.
The company said the shift moves enterprises beyond isolated hardware deployments towards production-ready AI factories, where performance, operational resilience, and AI governance are engineered across the entire lifecycle.
Alongside this, ASUS reaffirmed its unified AI infrastructure platform, which spans accelerated computing, networking, storage, deployment software, AI operations and governance.
By bringing infrastructure engineering, deployment and operational intelligence together into a single continuous platform, ASUS said it aims to help organisations shorten time to first token, accelerate time to revenue and improve infrastructure efficiency, while maintaining trusted AI at scale.
This positions ASUS as an end-to-end AI infrastructure partner, the company said, turning computing capacity into measurable business outcomes.
Commenting on the strategy, Paul Ju, Senior Vice President and Head of Infrastructure Solution Business Group at ASUS, said, “ASUS is now comprehensively embracing its All in on AI strategic initiative, dedicated to embedding AI across infrastructure, commercial solutions, and innovative physical devices.”
“As AI factories evolve into always-on production environments, infrastructure performance and AI governance can no longer be addressed separately. Infrastructure is the foundation of AI production itself. From the decisions that shape the first token to the controls that govern AI at scale, ASUS is engineering for what comes next—bringing compute, connectivity, power, cooling and automation together to turn every watt, every dollar and every minute invested into AI output”, Ju added.
Turning to governance, ASUS said trust in an AI factory has to be established well before any model goes live.
To that end, the company combines its system engineering approach with the NVIDIA DSX Sim Blueprint, allowing enterprises to design, simulate, validate and optimise complete AI factories before physical deployment begins.
Working alongside ecosystem partners including Schneider Electric, AVEVA and IBM, ASUS brings compute, networking, storage, power, cooling and facility infrastructure into a single digital twin environment, according to the company.
This, ASUS said, allows organisations to identify constraints, optimise configurations and reduce deployment risk before installation gets under way.
Once a system is deployed, governance continues throughout ongoing AI operations, the company noted.
As agentic and physical AI systems grow more autonomous, ASUS said organisations need continuous visibility, oversight and operational control across models, data, tools and enterprise systems.
Beyond its existing infrastructure management platforms — ACC (ASUS Control Centre) and AIDC (ASUS Infrastructure Deployment Centre) — ASUS said it is extending these capabilities through its AI Software Stack and Turnkey AI Applications.
Within the AI Software Stack, customers gain access to tools including Quota & Billing and an MLOps Portal, which the company said support more efficient resource management within the data centre and faster overall operations.
At the application level, ASUS described the AI Software Stack as an “operational harness” that connects enterprise governance directly with AI services and autonomous agents, enabling secure and production-ready deployment of agentic AI.
The company credited its integration capabilities and long-standing collaboration with NVIDIA for enabling enterprises to move faster from design to deployment and into ongoing operations.
ASUS pointed to business growth as evidence of rising demand for integrated AI infrastructure that shortens time to first token, accelerates time to revenue and supports AI factories, ranging from frontier-scale model development through to enterprise, agentic and edge AI deployments.
For large-scale AI factories, the company highlighted its ASUS AI POD, XA VR721-E3, built on the NVIDIA Vera Rubin NVL72 platform, which it said provides a rack-scale system for high-volume intelligence and lower-cost token generation, delivering ten times the performance per watt compared with the previous generation.
Complementing this, ASUS said its enterprise-ready XA NR1I-E12LR and XA NR1I-E12L systems, based on the NVIDIA HGX Rubin NVL8 platform, are designed to accelerate high-density AI infrastructure for demanding enterprise training, inference and post-training workloads.
For agentic AI applications, ASUS introduced the XA P2N-E2, a 2U NVIDIA MGX platform equipped with two NVIDIA Vera CPUs and support for up to two dual-slot NVIDIA GPUs, which the company said is built for responsive, energy-efficient agentic reasoning, data processing and orchestration at scale.
On edge and visual computing, ASUS pointed to its ESC8000-E12P system, which features NVIDIA RTX PRO 6000 and RTX PRO 4500 Blackwell Server Edition GPUs and extends acceleration for enterprise inference, vision AI and visual computing from the data centre to the edge.
In addition, the company’s compact PE3000N edge-AI platform, powered by the NVIDIA Jetson Thor T5000 module, is designed to deliver real-time inference, sensor fusion and autonomous control for robotics and industrial automation environments, according to ASUS.
On the data foundation side, ASUS highlighted its UF920-E3-RS24 server, built on the NVIDIA STX modular foundation for AI-native storage, alongside its OJ340A-RS60 object storage solution and VS320D-RS26N storage system.
Together, the company said, this portfolio is designed to meet the capacity, availability and data management demands of modern AI pipelines, supporting AI factories from design validation through to production-scale deployment.
Beyond infrastructure, ASUS also showcased partner demonstrations intended to illustrate real-world AI applications.
Foxlink demonstrated industrial automation using autonomous robotics, while Aleria highlighted what it described as the need for Sovereign AI deployments to support localised data control and secure operations, according to ASUS.
The company said this approach, combined with its software-hardware integration and Kubernetes platforms, is intended to shorten time-to-revenue for customers.
