Liquid-Cooled AI Infrastructure: Powering Scalable Enterprise Intelligence

Enterprise AI isn’t slowing down—and neither are the demands it’s putting on traditional infrastructure. As compute density rises, air-based cooling can’t keep pace. Sustained GPU workloads generate intense, concentrated heat—pushing energy use higher, driving up operational costs, and putting real pressure on efficiency measures like Power Usage Effectiveness (PUE).

Scaling up is a real challenge. As we add more AI capabilities, energy use and system complexity also increase. This can slow down innovation, especially when speed is most important.

Solving this problem requires more than incremental tweaks to airflow. It calls for a rethink of thermal management at the rack level. Liquid cooling efficiently removes heat at the source, enabling higher-density computing, improving energy efficiency, and supporting sustainable, long-term scalability.

For companies making AI a core part of their operations, thermal design is more than just a technical detail. When done well, it drives better performance and efficiency, and gives a real competitive edge.

Liquid Cooling: Enabling High-Density AI

As enterprises push AI workloads to unprecedented density, traditional air-based cooling alone can’t keep pace. That’s where liquid cooling steps in, replacing or supplementing the air methods by circulating thermally conductive fluids directly to the hottest components, removing heat more efficiently at the source. This stabilizes operating temperatures, reduces hardware stress, and enhances reliability, uptime, and system lifespan—essential for mission-critical AI and data center operations.

Beyond reliability, liquid cooling enables the scaling of high-density workloads. By supporting greater computational density, it allows enterprises and research initiatives to accelerate complex AI tasks without being constrained by heat or inefficiency. Some of the key benefits include:

• Optimized AI performance supports higher-speed, more complex workloads without compromise.
• Improved PUE reduces energy waste and increases operational efficiency.
• Lower operational costs minimize energy consumption and infrastructure strain.
• Safer, quieter environments create better working conditions for personnel and sensitive operations.

By tackling thermal bottlenecks right at the source, liquid cooling stops being just a supporting function. It becomes a foundation for scalable, energy-efficient, high-performance AI infrastructure.

ASUS Optimized Liquid-Cooling Solutions

ASUS offers a comprehensive portfolio of liquid-cooling technologies, designed to meet the thermal, density, and operational demands of high-performance AI and HPC workloads. Enabled through strategic collaborations with global infrastructure leaders, these solutions combine efficiency, reliability, and scalability:

• Direct-to-Chip Cooling (D2C): Cold plates are placed directly on CPUs and GPUs to boost performance, extend hardware lifespan, and reduce energy consumption; ideal for sustained, high-density workloads.
• Direct-to-Chip Cooling with In-Row CDU: Combines chip-level precision cooling with centralized row control to maximize thermal efficiency and simplify large-scale AI deployment.
• Liquid-to-Air Cooling: Absorbs system heat via liquid and dissipates it through air, providing a flexible upgrade path compatible with existing infrastructure.

ASUS has a variety of flagship systems that showcase the full potential of liquid cooling, such as:

• ASUS AI POD with NVIDIA Vera Rubin NVL72 – A 100% liquid-cooled, rack-scale powerhouse engineered for massive AI workloads, featuring NVIDIA Vera Rubin NVL72 GPUs. With up to 227 kW TDP, it is purpose-built to handle trillion-parameter AI models in large-scale AI factories.
• XA NR1I-E12L / E12LR – Hybrid and full liquid-cooled options, balancing D2C GPU cooling with air-cooled CPUs for seamless infrastructure transitions.
• HGX Rubin server series, XA NB3I-E12, ESC8000A-E13X, and ESC8000A-E13P RTX PRO™ servers – High-performance, scalable servers for extreme GPU connectivity, memory-intensive AI, and power-efficient AI/visual workloads.

Through strategic partnerships with NVIDIA, Vertiv, Schneider, and certified storage providers (IBM, DDN, Weka, VAST Data), ASUS delivers turnkey, fully integrated AI infrastructure spanning servers, storage, networking, and AI services.

Real-World Application: NCHC Nano4 — Taiwan’s Liquid-Cooled AI Supercomputer

The Nano4 AI supercomputer in Taiwan is a strong example of the real-world impact of ASUS liquid cooling solutions. Developed with the National Center for High-Performance Computing (NCHC), Nano4 was designed to meet extreme performance and energy-efficiency requirements, enabling national-scale AI and HPC workloads that conventional air cooling could not support. It demonstrates how ASUS liquid cooling drives peak AI performance, keeps operations running smoothly, and enables scalable infrastructure—connecting advanced research, national projects, and enterprise-grade AI deployments.

Key elements of the deployment include:

• High-Performance Architecture: Integration of NVIDIA GB200 NVL72 and HGX H200 GPU clusters in a dual-compute design supports dense AI workloads at scale.
• Thermal Management Excellence: Direct-to-Chip (D2C) liquid cooling efficiently manages intense heat, maintaining optimal temperatures and reducing hardware stress.
• Rapid, Turnkey Deployment: Using the ASUS Infrastructure Deployment Center (AIDC), setup time was reduced from three weeks to just three days, accelerating access to critical AI research resources.
• Unified, Reliable Infrastructure: High-reliability storage, validated networking, and production-ready systems ensure operational stability for generative AI workloads.

Turning Thermal Challenges into an AI Advantage

Liquid cooling is no longer just a technical upgrade—it's a strategic enabler of the next generation of AI and HPC innovation. By efficiently managing heat and supporting high-density compute, it lets enterprises hit peak performance, scale complex workloads, and adopt emerging AI technologies with confidence.

At ASUS, we provide turnkey, validated solutions and full-stack infrastructure expertise so organizations can move from experimental research to production-scale AI without the usual headaches. The result: future-proofed data centers, smoother operations, and drive measurable business value  for years to come.

Learn more about ASUS Liquid-Cooling Solutions.