ASUS has announced two NVIDIA HGX Rubin NVL8 server systems scheduled for launch in Q4 2026, offering hybrid and fully liquid-cooled configurations for AI data centers. Presented at NVIDIA GTC Berlin 2026 in Germany, the systems will be available in partial-rack and full-rack configurations to support large-scale AI workloads.
The ASUS XR R1I12B-M1BHFS features a 3U design combining direct-to-chip (D2C) liquid cooling for the NVIDIA HGX Rubin 8-GPU baseboard with air cooling for dual Intel Xeon 6 processors. According to ASUS, the air-cooled CPU sled reduces maintenance time by up to 50%. The system also supports flexible U.2 and E1.S storage configurations.
The XR R1I12B-M1BLFS uses a 2U, fully liquid-cooled architecture designed for high computing density, thermal efficiency, and improved power usage effectiveness (PUE). Its NVIDIA-validated design incorporates a cable-less board-to-board architecture intended to simplify system integration and deployment.
ASUS is also demonstrating AI factory design and simulation capabilities using NVIDIA DSX Sim Blueprint. Working with Schneider Electric, AVEVA, and IBM, the company integrates computing, networking, storage, power, cooling, and facility infrastructure into a unified digital twin environment. This approach enables customers to evaluate infrastructure configurations, identify constraints, and assess system performance and efficiency before deployment.
The new servers are part of a broader ASUS AI portfolio showcased at GTC Berlin 2026, including NVIDIA Grace Blackwell-powered systems, workstations, and edge computing devices.
