Data center liquid coolants

Direct-to-chip liquid cooling is a cooling method primarily used in data centers, in which liquid coolant circulates through cold plates attached directly to the main heat-generating components such as CPUs, GPUs and memory modules. The coolant absorbs heat at the source and carries it away from the server.

The heated coolant is then transported to a heat exchanger, where it is cooled before being recirculated through the system. This closed-loop process removes heat directly from the components that generate it, providing targeted thermal management in modern data center environments.

Direct-to-chip cooling removes heat directly from the main heat-generating components in the server, such as CPUs and GPUs. This significantly reduces reliance on ambient air and allows heat to be managed efficiently at component level.

Because heat is removed at the source, direct-to-chip systems can support much higher rack densities, often in the range of 20–50 kW per rack or more. They can be integrated into existing server hardware with relatively minor modifications, making them a practical solution for many high-performance environments.

Direct heat removal also reduces reliance on fans. Depending on the system design, fans can be made smaller or removed altogether, lowering noise levels and reducing the risk of mechanical failure inside the server.

Direct-to-chip cooling and immersion cooling are two different approaches to liquid cooling in data centers. Both use liquid to remove heat more efficiently than air-cooling systems, but they differ in how the coolant interacts with server hardware.

In direct-to-chip cooling, liquid coolant circulates through cold plates attached directly to heat-generating components such as CPUs and GPUs. The coolant absorbs heat at the source and carries it to a heat exchanger, where it is cooled and recirculated. This approach allows precise heat removal while keeping server hardware out of the coolant loop.

Immersion cooling takes a different approach. Instead of circulating coolant through cold plates, entire servers or components are submerged in a dielectric fluid that does not conduct electricity. The fluid absorbs heat directly from all submerged hardware before the heat is removed from the system.

Direct-to-chip cooling is typically used where targeted component cooling and compatibility with existing server architectures are important. Immersion cooling can offer extremely high heat-removal capacity, but usually requires specialised hardware and infrastructure.

Immersion cooling requires tanks or enclosures in which servers are submerged, taking up more floor space and adding significant weight. This can lead to more substantial changes to the data hall layout and supporting infrastructure. Direct-to-chip cooling can typically be integrated into standard rack-based server layouts with limited additional space, mainly using cold plates, tubing and manifolds.

In modern data centers, air cooling becomes increasingly difficult to scale. AI and high-performance computing (HPC) workloads raise computing density and heat output, often pushing airflow-based systems towards their practical limits.

In high-density server environments, air cooling can struggle to dissipate heat effectively. Uneven airflow may create localised hot spots that increase the risk of equipment failure. Air-based systems require substantial energy to operate fans, chillers and CRAC (Computer Room Air Conditioning) units. They typically demand larger physical infrastructure, such as raised floors and ducting.

Liquid cooling offers a more efficient alternative. Because liquids have much higher thermal conductivity and heat capacity than air, they can remove heat more effectively from critical components at the source. This enables higher rack densities, improved energy efficiency and more reliable thermal management in modern data center environments. By reducing reliance on energy-intensive air handling and cooling infrastructure, liquid cooling can lower Power Usage Effectiveness (PUE). The more direct and controlled heat capture also makes heat recovery easier to implement, creating opportunities to improve Energy Reuse Effectiveness (ERE).