Building AI factories at enterprise scale

Artificial Intelligence (AI) has emerged as a dominant force in the tech sector and beyond, with global server market growth reaching 30% in a single year, reaching $122.6bn in Q1 2025. IDC now views AI-driven investment as durable rather than cyclical, predicting inference to account for over 70% of computing demand by next year.

This surge in AI infrastructure investment is extending beyond hyperscalers to enterprise environments, as companies strive to run agentic AI in-house. Enterprises seek data control, predictable costing, and life cycle services, particularly in regulated sectors like finance, healthcare, and manufacturing.

ASUS, a major player in the industry, expects this AI boom to continue in datacenters, edge locations, and developers' desks. The company unveiled an all-encompassing AI solution called the Enterprise Agentic AI Solution at NVIDIA's GTC show in Berlin on Oct 21-22. This three-layer offering includes AI servers, software, and a services operation, aiming to cater to AI needs at various stages of the AI continuum.

ASUS AI servers and infrastructure comprise a selection of AI servers and new units, tailored to support complex task execution and large-scale AI agent deployment. ASUS AI Software Stack provides software for integrating and controlling AI workflows, while ASUS AI HUB offers a services operation that ties everything together. ASUS's senior vice president and head of infrastructure solution BG, Paul Ju, highlights the company's focus on expanding from GPUs to CPUs, as agentic AI workloads increasingly revolve around CPU-dependent continuous 'agent loops'.

The company's solutions at GTC include single and full-rack systems built around the NVIDIA HGX Rubin NVL8 eight-GPU server board, targeting large AI reasoning workloads. ASUS offers 3U hybrid-cooled and 2U 100% liquid-cooled systems, designed for different use cases and promoting serviceability to reduce downtime and lower total cost of ownership.