Direct Liquid Cooling & Prefabricated Data Centres
Enabling Sustainable High-Performance Compute Infrastructure for Defence, Research and Sovereign AI Applications
Executive Summary
The rapid growth of Artificial Intelligence (AI), Machine Learning (ML), High Performance Computing (HPC), Digital Engineering and Modelling & Simulation is driving unprecedented demand for computational infrastructure.
Traditional enterprise data centres, originally designed around air-cooled IT loads of between 5kW and 15kW per rack, are increasingly challenged by modern GPU-based systems capable of exceeding 100kW per rack and continuing to rise.
At the same time, organisations are facing increasing pressure to reduce energy consumption, improve sustainability, accelerate deployment timelines and maximise utilisation of existing estate.
As a result, Direct Liquid Cooling (DLC) and Prefabricated Data Centre Infrastructure are emerging as complementary technologies capable of addressing these challenges.
This paper explores the benefits of Direct Liquid Cooling and examines how the integration of DLC within prefabricated modular data centre environments can provide an effective pathway for delivering scalable, efficient and future-ready digital infrastructure for defence, research and scientific computing applications.
Introduction
The Compute Challenge
The modern digital landscape is increasingly dependent upon computational capability.
Across defence, government, academia and industry, advanced compute platforms are now supporting:
- Artificial Intelligence and Machine Learning
- Large Language Models (LLMs)
- Scientific Research
- Digital Engineering
- Modelling & Simulation
- Synthetic Training Environments
- Data Analytics
- Cyber Security Operations
- Autonomous Systems Development
These workloads are driving dramatic increases in compute density and associated thermal loads.
Historically, data centres relied upon air as the primary cooling medium. Whilst suitable for conventional enterprise environments, air cooling becomes increasingly inefficient as rack densities rise.
The challenge facing infrastructure designers is no longer simply how to deploy additional compute capacity, but how to do so in a manner that remains energy efficient, sustainable and economically viable.
The Limitations of Traditional Air-Cooled Data Centres
Traditional data centres were designed during a period when server power densities were comparatively low.
Whilst advances in airflow management, containment systems and free cooling technologies have improved efficiency, there are practical limits to what can be achieved using air alone.
As rack densities increase, operators often experience:
- Increased cooling energy consumption
- Higher fan power requirements
- Greater mechanical plant capacity
- Reduced space efficiency
- Escalating infrastructure costs
- Physical limitations within existing facilities
In many environments, cooling infrastructure is becoming the primary constraint on future compute growth.
Direct Liquid Cooling: A More Efficient Approach
Understanding Direct Liquid Cooling
Direct Liquid Cooling removes heat directly from key components such as CPUs, GPUs and memory through the use of liquid-cooled cold plates.
Rather than relying solely on air to transport heat away from IT equipment, heat is transferred directly into a liquid cooling circuit and subsequently rejected through a heat exchanger system.
This enables significantly more efficient thermal management.
The fundamental advantage is simple:
Water is substantially more effective at transferring heat than air.
By transporting heat directly from the source, Direct Liquid Cooling reduces the energy required to cool high-performance compute environments.
Key Benefits of Direct Liquid Cooling
Increased Compute Density
Direct Liquid Cooling enables significantly higher rack power densities than are typically achievable using air cooling alone.
Benefits include:
- Support for 80kW+ rack deployments
- Support for advanced GPU clusters
- Reduced data centre footprint
- Improved utilisation of available floor space
This makes DLC particularly attractive for AI and HPC environments where compute density is a key requirement.
Improved Energy Efficiency
Direct Liquid Cooling reduces the need for:
- High-speed server fans
- Large volumes of conditioned air
- Extensive mechanical cooling infrastructure
As a result, overall facility energy consumption can be reduced.
Lower cooling energy requirements contribute directly to improved operational efficiency and reduced carbon emissions.
Improved Power Usage Effectiveness (PUE)
Power Usage Effectiveness (PUE) remains one of the most widely recognised measures of data centre efficiency.
By reducing cooling-related energy consumption, Direct Liquid Cooling can contribute to lower facility PUE compared with conventional air-cooled environments.
This becomes increasingly significant as compute densities continue to rise.
Future Heat Recovery Potential
Liquid cooling systems operate using warmer return water temperatures than conventional chilled air environments.
This creates opportunities for:
- Building heating integration
- District heating schemes
- Energy reuse programmes
- Sustainability initiatives
Whilst not applicable to every deployment, heat recovery represents an increasingly attractive option for organisations seeking to maximise energy utilisation.
Direct Liquid Cooling in Defence and Research Environments
Defence and research organisations face a unique combination of challenges. Infrastructure must be capable of supporting:
- Rapidly evolving technology requirements
- Sensitive information processing
- Mission critical workloads
- Energy and sustainability targets
- Long-term scalability
At the same time, estate availability and programme timescales are often constrained.
This has led many organisations to consider alternative approaches to infrastructure delivery.
The Evolution of Prefabricated Data Centres
A Manufacturing-Led Approach
Prefabricated data centres adopt a fundamentally different delivery model to traditional construction.
Rather than building critical infrastructure entirely on site, major systems are designed, integrated and tested within a factory environment before being delivered for installation.
This approach is increasingly being adopted across both commercial and government sectors.
Benefits of Prefabricated Infrastructure
Accelerated Deployment
Traditional data centre projects can take many months or years to become operational.
Prefabricated infrastructure allows:
- Parallel construction activities
- Factory integration
- Pre-testing of systems
- Reduced site installation activities
This can significantly reduce overall deployment timelines.
Improved Quality Assurance
Factory environments provide:
- Controlled working conditions
- Repeatable manufacturing processes
- Consistent quality standards
- Integrated testing procedures
The ability to conduct comprehensive Factory Acceptance Testing prior to shipment reduces commissioning risk and improves deployment certainty.
Reduced Site Disruption
Compared with conventional construction, prefabricated infrastructure can significantly reduce:
- Site labour requirements
- Construction traffic
- Health and safety exposure
- Environmental disruption
This can be particularly advantageous on operational defence, research and scientific sites.
Scalability and Flexibility
Modular architectures support a "build-as-required" approach.
Additional capacity can be introduced incrementally, allowing organisations to align infrastructure investment with demand.
This approach reduces over-provisioning and supports long-term flexibility.
The Combined Benefit of Direct Liquid Cooling and Prefabricated Data Centres
The greatest benefits are achieved when Direct Liquid Cooling and Prefabricated Infrastructure are considered together.
Rather than retrofitting cooling technologies into existing facilities, power, cooling and IT systems can be designed as an integrated solution from inception.
This approach enables:
Optimised Thermal Design
Cooling systems can be engineered specifically around expected compute densities.
Reduced Energy Consumption
Integrated power and cooling systems can be optimised to minimise losses and maximise efficiency.
Accelerated Delivery
Factory-based integration significantly reduces programme duration.
Improved Predictability
Testing and validation occur before equipment arrives on site.
Future Scalability
Additional modules can be deployed as compute requirements evolve.
Sovereign AI, National Security and Research Capability
A Strategic Infrastructure Challenge
Artificial Intelligence is rapidly becoming a strategic capability.
Governments and research organisations increasingly recognise that access to compute infrastructure is becoming as important as access to data itself.
Future capability development across defence and national security sectors will depend upon the availability of secure, scalable and energy-efficient compute environments.
Applications include:
- Advanced AI model training
- Digital battlefield simulation
- Synthetic environments
- Intelligence analytics
- Autonomous systems development
- Electronic warfare modelling
- Scientific and engineering research
These applications require infrastructure capable of supporting increasingly power-dense compute platforms.
Reducing Dependence on Legacy Infrastructure
Many existing facilities were never designed to support modern AI workloads.
Attempting to retrofit high-density compute into ageing infrastructure can introduce:
- Increased costs
- Extended delivery timelines
- Energy inefficiencies
- Operational disruption
Purpose-designed liquid-cooled modular infrastructure provides an alternative route that enables organisations to deploy capability rapidly whilst maintaining operational continuity.
Accelerating Capability Deployment
In strategic environments, the speed at which computational capability can be deployed may directly impact organisational effectiveness.
Prefabricated infrastructure enables:
- Faster delivery programmes
- Predictable deployment schedules
- Reduced construction risk
- Earlier operational availability
This can support the rapid introduction of emerging technologies and research programmes.
Supporting Sustainability Objectives
Government organisations are increasingly expected to balance operational requirements with environmental commitments.
Direct Liquid Cooling and modular infrastructure can contribute towards:
- Reduced energy consumption
- Improved infrastructure utilisation
- Reduced embodied carbon through factory manufacture
- Improved opportunities for future heat recovery
These factors support both operational and sustainability objectives.
Looking Ahead
The continued growth of AI, HPC and advanced analytics will place increasing demands on data centre infrastructure.
Higher rack densities, increased power consumption and more sophisticated workloads will continue to challenge conventional cooling approaches.
Direct Liquid Cooling provides a practical and proven mechanism for addressing these challenges, whilst prefabricated infrastructure offers a scalable and efficient deployment model.
Together, these technologies provide a framework for delivering future-ready digital infrastructure capable of supporting the next generation of defence, research and scientific computing applications.
Conclusion
The growth of AI and High-Performance Computing is fundamentally changing infrastructure requirements.
Traditional air-cooled environments are increasingly challenged by the thermal and power demands of modern compute platforms.
Direct Liquid Cooling offers a highly efficient method of supporting high-density compute environments whilst reducing cooling energy requirements and improving overall facility performance.
When combined with prefabricated modular data centre infrastructure, organisations can also realise significant benefits in deployment speed, scalability, quality assurance and operational flexibility.
For defence, research and scientific organisations seeking to develop future computational capability, the convergence of Direct Liquid Cooling and Prefabricated Data Centre Infrastructure provides a compelling pathway towards sustainable, scalable and resilient digital infrastructure.
As demand for sovereign AI capability continues to grow, these technologies are likely to play an increasingly important role in enabling secure, high-performance compute environments capable of supporting future mission requirements.
Return to Knowledge Hub