2.3kW
Extreme GPU Power
Stay Cool, Scale Fast
with NVIDIA HGX™ Rubin 8-GPU AI server
In the fierce race for next-gen AI infrastructure, efficiency, density, and reliability are now must-haves. ASUS delivers the optimized NVIDIA HGX™ architecture with the fully liquid-cooled, and hybrid-cooled rack solution powered by HGX™ Rubin—built for massive reasoning and generative AI. Liquid- or hybird-cooled, it ensures high-volume production, on-time delivery, lower TCO, and unwavering performance. Discover how ASUS + NVIDIA turns AI ambition into reality.
Flexible & Powerful Cooling
Your Choice of Hybrid or 100% Liquid Cooling
Addressing the most rigorous demands of the modern data center, ASUS proudly introduces the fully liquid-cooled, and hybrid-cooled rack solution built on the NVIDIA HGX™ Rubin platform. To ensure a seamless and cost-effective transition to liquid cooling, ASUS offers two distinct solutions tailored to your infrastructure needs.
XR R1I12B-M1BHFS
Hybrid-Cooled Rack Solution
with 3U Hybrid-Cooled Compute Trays
This innovative design combines direct-to-chip (D2C) liquid cooling for the high-performance NVIDIA HGX™ Rubin 8-GPU baseboard with efficient air cooling for the dual Intel® Xeon® 6 processors. It offers a perfect balance of performance and ease of adoption.
XR R1I12B-M1BLFS
Fully Liquid-Cooled Rack Solution
with 2U Fully Liquid-Cooled Compute Trays
Designed for maximum thermal efficiency and density, this 100% liquid-cooled system is the ultimate solution for handling the most intensive AI workloads in high-density environments.
The Engine of AI Innovation, NVIDIA HGX™ Rubin NVL8
Engineered for massive-scale AI model training, ASUS fully liquid-cooled, and hybrid-cooled rack solution delivers a quantum leap in performance for training and inference.
800G
Ultra-fast GPU Bandwidth
Modular GPU Sled
Built for Serviceability
Maximize Uptime with Modular Architecture and Rapid Serviceability
The modular design philosophy of ASUS fully liquid-cooled, and hybrid-cooled rack solution is centered on minimizing downtime. Components like the CPU sled and NVIDIA Rubin GPU sled are designed for tool-less access, allowing for faster maintenance and upgrades. This focus on rapid serviceability decreases operational interruptions and lowers the total cost of ownership.
Modular CPU Assembly
Independent GPU Sled
A Complete Rack-Scale Reference Architecture with NVIDIA HGX™ Rubin Platform
Scale from a single server to a fully integrated AI powerhouse with the ASUS AI Factory reference architecture. This pre-validated configuration ensures rapid deployment and optimized performance at the rack level.
-
-
2 x North-South Networking-NVIDIA SN2201
1 x North-South Networking-NVIDIA SN5610
23 x Management Node-RS700-E12-RS4U
8 x File System-RS501A-E12-RS16U
1 x Object Storage-OJ340A-RS60
2x 3RU Power Shelf 110kW
1 x Management switch - NVIDIA SN2201
2 x 3U Power Shelves 110k
9 x ASUS XANR1I-E12LR
• 18 x Intel® Xeon® 6 processors
• 72 x NVIDIA Rubin GPUs2 x 3U Power Shelves 110k
-
Building the Next Frontier of AI
From a single system to an entire rack, ASUS provides the foundational building blocks for your AI ambitions. Our next-generation cooling solutions with NVIDIA HGX™ Rubin platform reduce deployment effort and empower you to build trusted, powerful, and scalable AI infrastructure for the future.
FAQ
-
What is the difference between hybrid and fully liquid-cooled HGX™ Rubin servers?What is the difference between hybrid and fully liquid-cooled HGX Rubin servers? • Hybrid-Cooled Compute Tray: Combines liquid cooling for the NVIDIA HGX™ 8-GPU baseboard with air cooling for the CPUs. It's ideal for an easy transition to liquid cooling.
• Fully Liquid-Cooled Compute Tray: Provides 100% liquid cooling for maximum thermal performance and density in demanding data centers. -
What are the main benefits of liquid cooling for an NVIDIA HGX™ Rubin 8-GPU AI server?What are the main benefits of liquid cooling for an NVIDIA HGX Rubin 8-GPU AI server? Liquid cooling is essential for next-generation AI servers. Key benefits include:
• Peak Performance: Prevents thermal throttling of powerful GPUs.
• Higher Density: Allows more servers per rack, maximizing compute power.
• Greater Efficiency: Lowers data center energy consumption and operational costs. -
How does the NVIDIA HGX™ Rubin platform enhance AI performance?How does the NVIDIA HGX Rubin platform enhance AI performance? The NVIDIA HGX™ Rubin platform accelerates AI workloads by providing:
• Extreme Power: Manages the high thermal demands of the latest GPUs.
• Massive Bandwidth: Offers 800G of integrated bandwidth per GPU to eliminate data bottlenecks.
• High Availability: Features a modular sled design for faster service and maximum uptime. -
What is the reference architecture of ASUS AI POD with NVIDIA HGX™ Rubin platform?What is the reference architecture of ASUS AI POD with NVIDIA HGX™ Rubin platform? The ASUS AI Factory is a pre-validated, rack-scale blueprint that integrates multiple ASUS HGX™ Rubin servers with all necessary networking, power, and storage. It provides a proven path to rapidly deploy a complete AI supercomputer, scaling from a single server to a full rack with confidence.
-
How does the server’s modular design improve serviceability and reduce downtime?How does the server’s modular design improve serviceability and reduce downtime? The server features a modular design with tool-less access to critical components like the CPU and GPU sleds. This allows for significantly faster maintenance and upgrades, which directly reduces operational downtime and lowers the total cost of ownership (TCO).