As generative AI and high-performance computing continue to scale, the pressure on data centre infrastructure is no longer incremental, it’s exponential.

More compute is being deployed into the same physical footprint. Power densities are increasing. Expectations around efficiency, availability, and scalability are rising in parallel.

Cooling is no longer a supporting function. It is becoming a defining factor in how data centres are designed, expanded, and operated.

At the centre of this shift is Direct Liquid Cooling (DLC).

The Constraint Isn’t Just Cooling - It’s Space

Introducing higher cooling capacity has traditionally meant introducing more infrastructure into the data hall.

In practice, this creates a trade-off. Additional systems, whether in-row units, containment, or supplementary air cooling, occupy space that would otherwise be allocated to IT racks.

As density increases, that trade-off becomes more pronounced. Direct Liquid Cooling changes how that capacity is deployed.

Rather than concentrating infrastructure within the white or grey space depending on your setup, DLC enables core systems, such as Coolant Distribution Units (CDUs), manifolds, and distribution pipework, to be positioned externally in grey space areas like mechanical corridors and service zones.

Cooling is delivered directly to the rack, while the infrastructure that supports it is removed from the data hall environment. The result is the ability to introduce higher-density cooling without reducing available rack footprint.

Cooling at the Source

The shift to liquid cooling is driven by physics as much as infrastructure.

Air cooling is reaching its limits in high-density environments. Liquid, by contrast, has a significantly higher heat capacity. Direct Liquid Cooling leverages this by directing coolant to the point where heat is generated,  at the chip level.

This approach allows greater thermal loads to be managed more efficiently, supporting the requirements of AI and HPC workloads where heat output is concentrated and continuous.

Integrating Into Existing Environments

For many facilities, the challenge is not building new capacity, but adapting what already exists.

Rittal’s Direct Liquid Cooling systems are designed to integrate into existing chilled water loops using liquid-to-liquid configurations. This allows new cooling capability to be introduced without the need for additional cooling towers or major plant infrastructure.

By working within the existing mechanical framework, implementation can be achieved without extensive structural modification, while also avoiding additional mechanical load on building systems.

At the same time, positioning infrastructure in grey space supports installation and maintenance without direct interaction with active IT environments.

Designed for Density, Built for Scale

As workloads increase, cooling systems must scale in step.

Rittal’s DLC portfolio is structured around a modular architecture, with solutions ranging from 70 kW through to 1 MW. This includes:

• Rear Door Heat Exchangers (70 kW) for rack-level cooling
• In-Rack CDUs (150 kW) for targeted deployments
• In-Row CDUs (up to 1 MW) for high-density environments

This modular approach allows capacity to be introduced incrementally, aligning with demand rather than requiring full-scale infrastructure changes upfront.

Reliability in Live Environments

In high-availability data centres, system design must account for continuous operation.

Rittal’s Direct Liquid Cooling systems incorporate redundancy through configurations such as N+1 pump arrangements, ensuring cooling capacity is maintained in the event of a component failure. Hot-swappable components allow critical elements, such as pumps, sensors, and controllers, to be replaced while the system remains operational. This enables maintenance without downtime and supports consistent cooling performance in environments where interruption is not an option.

Additional features, including leak monitoring and integrated control systems, further support long-term operational reliability.

A Platform for AI Infrastructure

Beyond individual components, Direct Liquid Cooling is increasingly being deployed as part of a broader platform approach.

Rittal’s solutions are designed to align with OCP architectures, supporting standardisation across data centre environments. This includes integrated power distribution via DC busbar systems and configurable manifold designs for efficient coolant distribution.

The result is a system that supports not just cooling, but the broader requirements of hyperscale and AI infrastructure, efficiency, scalability, and repeatability.

Rethinking How Capacity Is Delivered

As AI workloads continue to drive density, the challenge is no longer simply how to remove heat. It is how to do so within the constraints of existing environments, without compromising available IT capacity.

Direct Liquid Cooling addresses both.

By removing supporting infrastructure from the data hall and relocating it into grey space, DLC allows cooling capacity to scale without encroaching on white or grey space, depending on your setup. At the same time, its ability to integrate with existing systems and operate within live environments makes it suited to both new builds and retrofit scenarios.

In this context, cooling is no longer just a technical requirement. It is a key enabler of how data centres grow, adapt, and remain viable in the face of increasing demand.