In conjunction with the news of joining the NVIDIA GTC 2026 expo, ASUS has unveiled its ASUS Optimized Liquid-Cooling Solutions alongside a strategic partner framework designed to tackle one of the biggest problems in next-gen AI and HPC data centers: heat, power, and density going through the roof.
This new approach is apparently designed for the upcoming SKUs of ASUS AI PODs based on the NVIDIA Vera Rubin NVL72 design to deliver even higher heat removal effectiveness and efficiency from denser racks of accelerators and interconnecting wires, resulting in lower energy consumption, PUE (1.18), and TCO.
However, this isn’t a one-size-fits-all approach because ASUS is also doing direct-to-chip (D2C) cooling, in-row CDU-based setups, and hybrid configurations, and through the power of partnerships like Schneider Electric and Vertiv for infrastructure collaboration as well as integrating precision components from players such as Auras Technology and Cooler Master, the whole solution will become a purpose-built, scalable cooling systems that can handle extreme rack density without compromising stability.
One such early deployment is done with the National Center for High-performance Computing under the National Institute of Applied Research in Taiwan, where the system features a dual-compute architecture combining the NVIDIA HGX H200 cluster and the NVIDIA GB200 NVL72 – marking Taiwan’s first fully liquid-cooled AI supercomputer deployment of this architecture.
GTC 2026 Info
ASUS will be running a big demo booth at GTC 2026 San Jose from March 16 – 19, with the booth number #421.
