At SC 2025, ASUS outlined its approach to next-generation AI infrastructure with hardware designed for large training clusters and enterprise deployment. The company showed systems built for a wide range of compute needs, from rack-scale supercomputing to servers suited for AI development within enterprises.
ASUS demonstrated platforms powered by NVIDIA technologies, including the XA GB721-E2 with the GB300 NVL72 configuration and the XA NB3I-E12 built on the HGX B300 platform. The ESC8000A-E13X server supports ConnectX-8 SuperNICs and is aimed at multi-node AI workloads that need steady data movement.
The company also presented the XA AM3A-E13, which uses AMD Instinct MI355X GPUs and EPYC 9005 processors. This setup focuses on high-density compute and power-efficient performance for applications such as generative AI. Another system, the ESC8000-E12P, uses Intel Xeon 6 processors with Gaudi 3 accelerators to provide balanced performance suited for AI, HPC and mixed enterprise workloads.
Alongside its hardware, ASUS highlighted its tools for storage, deployment and lifecycle management.
Leave a comment