ASUS has unveiled its new artificial intelligence infrastructure at GTC 2026, highlighting the ASUS AI POD as the centerpiece. This solution, based on the NVIDIA Vera Rubin platform, offers a comprehensive end-to-end system, fully liquid-cooled, aimed at maximizing performance and energy efficiency in enterprise and data center environments. Under the slogan "Reliable AI, Total Flexibility," ASUS proposes an adaptable framework that encompasses large-scale AI factories, desktop supercomputing, edge AI, and enterprise solutions, enabling the creation of high-performance, low-power AI clusters. The goal is to drastically reduce both PUE (Power Usage Effectiveness) and TCO (Total Cost of Ownership). As a provider of NVIDIA GB300 NVL72 and HGX B300 systems, ASUS positions the AI POD as its flagship solution: a rack-scale, liquid-cooled system designed for intensive AI workloads. Thanks to partnerships with leaders in cooling and components, ASUS offers customized cooling and redundancy options, adapting to any business need. The company backs its offering with specialized consulting, a wide range of AI and storage solutions, application integration, and ongoing services, ensuring scalability and sustainability. The flagship, XA VR721-E3, is based on the Vera Rubin NVL72 platform and is designed for AI models with trillions of parameters, delivering massive performance in large-scale AI environments. This fully liquid-cooled system achieves a TDP of up to 227 kW (MaxP) or 187 kW (MaxQ), achieving up to 10 times more performance per watt. To meet the demands of data centers, ASUS has launched servers based on NVIDIA HGX Rubin NVL8, with eight NVIDIA Rubin GPUs connected via sixth-generation NVLink and a bandwidth of 800G per GPU. To facilitate the transition to liquid cooling, ASUS offers two alternatives: the XA NR1I-E12L, with hybrid cooling (direct liquid to the chip for the motherboard and air for Intel Xeon 6 processors), and the XA NR1I-E12LR, fully liquid-cooled. The range is completed with high-performance servers such as the XA NB3I-E12 (based on HGX B300), the ESC8000A-E13X (with NVIDIA MGX and ConnectX-8 SuperNIC for extreme GPU connectivity), and the ESC8000A-E13P, equipped with NVIDIA RTX PRO 4500/6000 Blackwell Server Edition GPUs, ideal for AI workloads, data processing, and visual computing in an efficient format. ASUS covers the entire cycle of physical AI, from development to implementation. The ExpertCenter Pro ET900N G3, a desktop supercomputer with NVIDIA Grace Blackwell Ultra and 748 GB of unified memory, enables the training of massive models. The compact ASUS Ascent GX10, with the Grace Blackwell Superchip, offers petaflop-scale performance for rapid iterations and edge applications. The PE3000N, powered by NVIDIA Jetson Thor, provides 2070 TFLOPS for real-time inference, ideal for sensor fusion and autonomous navigation. These systems enable a unified workflow, integrating open models like NVIDIA Cosmos and Metropolis vision libraries for intelligent physical applications. To accelerate enterprise AI, ASUS has launched the AI Hub, a turnkey local platform based on ESC8000 servers and open-source LLM technology like NVIDIA Nemotron and Gemma. It allows the creation of custom AI assistants, implementation of document intelligence with RAG, and maintenance of data sovereignty. Internally tested with over 10,000 employees, the platform has demonstrated efficiency and accuracy, facilitating access to critical information through intelligent agents and natural language queries. ASUS and NVIDIA also collaborate on NVIDIA NemoClaw, an open-source stack that simplifies the secure and efficient execution of OpenClaw assistants. Sustainability is key to ASUS's strategy, integrating green computing innovations in hardware and software to minimize TCO and environmental impact. The servers incorporate Thermal Radar 2.0, with up to 56 sensors to optimize fan performance and reduce energy consumption by up to 36%. In software, the ASUS Control Center Data Center Edition centralizes management, enhances security, and automatically monitors carbon emissions, helping companies meet their ESG goals. Source: muycomputerpro.com