Building AI factories at enterprise scale
SPONSORED FEATURE: ASUS wants to scale AI everywhere in the enterprise
As the AI sector continues to expand, enterprises are increasingly building their own AI infrastructure to control data and secure predictable costs. With the global server market growing by 30% in the first quarter of 2025, IDC now sees AI-driven investments as durable. By the following year, inference is expected to make up over 70% of computing demand, with edge locations accounting for the fastest growth.
ASUS, a major player in the AI hardware space, is looking to meet this demand by offering a full-stack AI solution for enterprises at various stages of the AI continuum, including datacenters, edge locations, and developers' desks.
ASUS's offering includes AI servers, an AI software stack, and an AI hub that ties everything together. The company's AI servers are designed to support agentic AI, which has become increasingly important as compute demand grows. Unlike traditional AI servers that focus on GPU compute, ASUS's servers are optimized for CPU-dependent workloads, such as task planning, code execution, external tool invocation, and reflection or correction work.
This shift in focus has created bottlenecks in conventional AI servers due to single-thread performance limitations, memory bandwidth constraints, and latency issues caused by the multi-die design of CPUs.
ASUS aims to address these challenges by offering different AI server options tailored to various use cases. At the datacenter, ASUS has introduced both single and full-rack systems built around the NVIDIA HGX Rubin NVL8 eight-GPU server board. These systems cater to large AI reasoning workloads. The 3U hybrid-cooled system uses direct-to-chip liquid cooling and air-cooled dual Xeon 6 chips, providing a smooth transition into AI factory hardware.
In contrast, the 2U unit features 100% liquid cooling, a cable-free board-to-board design, and offers higher compute density and PUE. ASUS emphasizes the serviceability of these units, claiming up to a 50% time-saving on maintenance operations compared to other AI servers.
Written by urgent.news from The Register's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
Also reported by 1 other outlet
- Building AI factories at enterprise scale theregister.com
