ASUS AI Tech 2026 Seoul: Building the Infrastructure for the Next Era of AI
ASUS and partners at AI Tech 2026 Seoul outlined how evolving AI workloads demand integrated infrastructure beyond raw compute, emphasizing power, cooling, connectivity, and automation as interconnected priorities.
At ASUS AI Tech 2026 in Seoul, ASUS and ecosystem partners discussed how the transition from generative AI to agentic and physical AI is reshaping infrastructure needs. The event focused on technologies and partnerships required to support increasingly diverse and complex AI workloads efficiently and at scale. CH Hsieh, ASUS Corporate Vice President of R&D Center, stressed that infrastructure must evolve alongside AI models, addressing performance, efficiency, scalability, and governance. He highlighted a shift in industry focus from building larger data centers to optimizing the cost efficiency of token production.
ASUS emphasized that power, cooling, connectivity, and automation should be treated as interconnected components of modern AI infrastructure. The company argued that reducing inefficiencies across these elements improves performance, resource optimization, and the return on AI investments. CH Hsieh summarized this approach with the principle: 'Every Signal. Every Watt. Every Degree. Every Second,' reflecting the interconnected factors determining AI system efficiency and scalability.
The event showcased enterprise AI server platforms, HGX training systems, and rack-scale ASUS AI POD deployments powered by the NVIDIA Vera Rubin NVL72 server platform. These solutions are designed to support large-scale AI training and supercomputing workloads, illustrating ASUS's strategy for scalable AI infrastructure. The concept of 'AI factories' emerged as a key theme, representing integrated environments combining computing, storage, networking, deployment software, AI operations, and governance.
ASUS presented a portfolio spanning enterprise AI servers, advanced training platforms, and rack-scale AI POD deployments to meet evolving workload demands. The company highlighted solutions tailored for enterprise inference, model training, and large-scale AI environments, aiming to support AI adoption from individual workloads to enterprise-wide deployment. The event underscored that AI infrastructure is defined by more than compute power, requiring attention to power efficiency, cooling, connectivity, and automation to maximize performance.