What Is Liquid Cooling in Data Centers?

9.17.2026

Artificial intelligence and high-performance computing are changing how data centers manage heat. As CPUs and GPUs become more powerful and rack densities increase, traditional air cooling can become less practical for removing concentrated heat loads. ASHRAE has identified increasing processor power, cooling demands and airflow requirements as factors contributing to the expansion of liquid cooling in data centers.

Liquid cooling in data centers is a method of removing heat from servers by using a liquid coolant to capture and transport heat from high-power components. It can supplement or, in some architectures, significantly reduce reliance on traditional air cooling.

How Does Liquid Cooling Work in a Data Center?

One of the most common approaches is direct-to-chip cooling.

In a direct-to-chip system, a metal cold plate is mounted directly to a CPU, GPU or other high-heat component. Coolant flows through sealed channels inside the cold plate, absorbs heat from the processor and carries it away from the server.

In many direct-to-chip systems, the warmed coolant travels through the technology cooling system (TCS) to a coolant distribution unit (CDU). Pumps maintain the required flow, while a heat exchanger within the CDU transfers heat from the technology cooling system to the facility cooling system (FCS). ASHRAE provides additional information on design considerations for CDU implementations and non-CDU implementations as well as the distinction between TCS and FCS loops in its whitepaper.

From there, the heat can be rejected through equipment such as chillers, dry coolers or cooling towers before cooled liquid is circulated back through the system.

In simple terms:
CPU/GPU → cold plate → technology cooling loop → CDU/heat exchanger → facility cooling system → heat rejection

What Are the Main Types of Data Center Liquid Cooling?

Direct-to-Chip Cooling

Direct-to-chip cooling uses cold plates attached directly to high-heat components such as CPUs and GPUs. Coolant circulates through the cold plates without normally coming into contact with the electronics themselves.This approach is particularly relevant for high-density AI and high-performance computing environments.

Immersion Cooling

With immersion cooling, servers or components are submerged in a specially designed dielectric fluid that does not conduct electricity. The fluid absorbs heat directly from the hardware before transferring it to the cooling system.

Hybrid Cooling

Liquid cooling does not necessarily eliminate air cooling. Many data centers use a hybrid cooling approach, where liquid handles the highest heat loads while air continues to cool components that are not connected to the liquid loop.

Why Is Liquid Cooling Important for AI Data Centers?

AI workloads rely heavily on powerful GPUs and other processors that can generate significant amounts of heat in a relatively small space.

As more computing power is concentrated within each rack, moving enough air through the equipment becomes increasingly challenging. Liquid cooling captures heat closer to where it is generated, helping data centers accommodate higher-density computing environments. ASHRAE’s white paper on liquid cooling in mainstream data centers provides additional background on the thermal challenges associated with increasing equipment power and airflow requirements.

Depending on the overall system design, liquid cooling can also reduce reliance on server fans and create opportunities for more efficient thermal management.

What Role Do Pumps and CDUs Play in Liquid Cooling?

Reliable coolant circulation is essential to the liquid cooling process.

Pumps maintain circulation within the cooling loops that connect servers, CDUs, heat exchangers and facility cooling equipment. Depending on the system architecture, separate pumping systems may serve the technology cooling system and facility cooling system.

A CDU helps manage coolant flow and temperature within the technology cooling system and can use a heat exchanger to transfer heat to the facility cooling system.

For mission-critical data centers, pump selection, controls, monitoring and redundancy are important considerations when designing the complete cooling system.

As AI and high-performance computing continue to increase data center heat loads, efficient and reliable coolant circulation will remain an important part of keeping critical infrastructure operating.

Does Liquid Cooling Mean Water Touches the Electronics?

Usually, no.

In direct-to-chip systems, coolant remains contained within sealed cold plates, tubing, manifolds and other components of the cooling loop, so it does not normally come into direct contact with the electronics.

Immersion cooling is different because the hardware is intentionally surrounded by a specially selected dielectric fluid.
Ultimately, liquid cooling is a heat-transport system: cold plates or dielectric fluids capture heat, pumps circulate the coolant, CDUs can help manage and transfer heat between cooling loops, and the facility cooling system ultimately removes that heat from the data center.

Wilo USA and Brands Data Center Solutions

As data center heat densities increase, selecting the right cooling architecture and the pumps that support it is critical to maintaining reliable thermal performance. Wilo USA and Brands offer pump solutions for liquid cooling and facility cooling systems in data center applications.