Liquid-Cooled Container Data Center Guide: Design, CDU & Deployment | 2026

As AI training, GPU clusters, high-performance computing (HPC), cloud services, and high-density computing continue to expand, conventional air-cooled data centers are reaching practical limits. Higher rack power density increases cooling demand, raises HVAC energy consumption, and makes traditional data center construction slower and more expensive.

A liquid-cooled container data center offers a faster and more scalable alternative. By integrating liquid cooling infrastructure, server racks, cooling distribution units (CDUs), power distribution, monitoring, fire protection, and containerized enclosure systems into a prefabricated module, organizations can deploy high-density computing capacity in a shorter time frame.

This guide explains how a liquid-cooled container data center works, what equipment it includes, how to choose between cold plate and immersion cooling, and what to consider before deployment.

What Is a Liquid-Cooled Container Data Center?

A liquid-cooled container data center is a modular data center solution built inside a containerized enclosure. It combines IT infrastructure and thermal management systems into a factory-integrated unit that can be transported, installed, connected, and commissioned at the deployment site.

Unlike conventional data center construction, which often requires extensive civil works, mechanical installation, electrical installation, and on-site integration, a containerized liquid cooling solution can be pre-engineered and tested before shipment.

A typical liquid-cooled container data center may include:

Liquid-cooled server racks or immersion cooling tanks

Rack-level, cabinet-level, or container-level CDUs

Primary and secondary cooling loops

Dry coolers, cooling towers, chillers, or hybrid cooling systems

UPS systems, PDUs, switchboards, and busbar systems

Leak detection, temperature monitoring, flow monitoring, and pressure monitoring

Fire suppression, lighting, access control, and remote monitoring systems

Insulated, weather-resistant, and corrosion-resistant container structures

The purpose is not simply to place servers inside a container. It is to create an integrated high-density computing environment that can support faster deployment, better thermal efficiency, and modular expansion.

Why Liquid Cooling Is Becoming Essential for High-Density Computing

Traditional air cooling remains suitable for many low- and medium-density IT environments. However, as rack loads increase from 10 kW to 30 kW, 50 kW, 80 kW, or even higher, air-based cooling systems become increasingly difficult to scale efficiently.

AI servers and GPU clusters generate concentrated heat from processors, accelerators, memory modules, networking components, and power systems. Air has limited heat-transfer capability compared with liquid. As a result, high-density systems often require larger air-conditioning capacity, stronger airflow, more fan power, and more complex hot-aisle or cold-aisle containment.

Liquid cooling transfers heat more directly from the heat source to the cooling loop. This allows the data center to support higher rack density while reducing reliance on large air-handling systems.

Key advantages of a liquid-cooled container data center include:

Higher rack power density

Reduced cooling energy consumption

Faster deployment compared with conventional data center construction

Smaller physical footprint for the same IT capacity

Better stability for high-performance servers

Greater adaptability in hot, dusty, remote, or industrial environments

Easier modular expansion for phased projects

For AI data centers, HPC clusters, edge computing sites, and high-density mining operations, containerized liquid cooling can provide a practical balance between performance, construction speed, and operational efficiency.

How Does a Liquid-Cooled Container Data Center Work?

The basic principle is straightforward: liquid absorbs heat from IT equipment, transfers that heat through a cooling loop, and releases it through an external heat-rejection system.

Most liquid-cooled container data centers use two main cooling circuits: a primary loop and a secondary loop.

Primary Cooling Loop

The primary loop connects the containerized data center to an external heat-rejection system. Depending on the project location, climate, water availability, and energy targets, the external cooling source may include:

Dry coolers

Cooling towers

Chillers

Plate heat exchangers

Natural cooling systems

Hybrid cooling systems

The primary loop removes heat from the liquid cooling system and rejects it to the outside environment.

For example, a dry cooler may be suitable for locations with moderate or cool ambient temperatures and limited water availability. In hotter climates, a chiller or hybrid cooling system may be required to maintain stable supply-water temperatures during peak conditions.

Secondary Cooling Loop

The secondary loop serves the IT equipment directly. It delivers cooling liquid from the CDU to liquid-cooled server racks, cold plates, manifolds, or immersion tanks.

The CDU separates the primary and secondary loops through a heat exchanger. It also manages flow rate, pressure, temperature, pump operation, alarms, and system protection.

Important secondary-loop parameters include:

Supply liquid temperature

Return liquid temperature

Flow rate

Pressure

Pressure drop

Coolant cleanliness

Water quality or coolant condition

Leak detection and protection logic

A well-designed secondary loop is essential for stable operation. Even when the CDU has sufficient cooling capacity, poor hydraulic design, incorrect manifold sizing, or excessive pressure drop can affect server cooling performance.

Core Components of a Liquid-Cooled Container Data Center

Liquid-Cooled Server Racks

Liquid-cooled racks are designed to support high-density servers and cooling distribution systems. Depending on the application, they may use standard 19-inch racks, 21-inch OCP racks, or customized high-load server cabinets.

A high-density liquid-cooled rack may include:

High static load capacity

Supply and return liquid manifolds

Quick disconnect couplings

Leak detection sensors

Separate routing for power, network, and liquid lines

Flexible hose connections

Compatibility with AI servers, GPU servers, HPC servers, or hydro miners

For projects with rack loads above 40 kW, rack-level flow distribution and supply-return temperature control become particularly important.

Cooling Distribution Unit (CDU)

The CDU is one of the most important components in a liquid cooling system. It transfers heat between the primary and secondary cooling loops while controlling the liquid delivered to IT equipment.

CDUs can be configured as:

Rack-mounted CDUs

Cabinet CDUs

In-row CDUs

Container-level CDUs

Modular parallel CDUs

When selecting a CDU, key factors include cooling capacity, pump configuration, redundancy level, heat exchanger performance, interface size, control system, maintenance access, and compatibility with the selected server platform.

For mission-critical projects, common design options include N+1 pump redundancy, dual power supply, dual-loop architecture, remote monitoring, and automatic alarm response.

External Heat-Rejection Equipment

The external cooling system determines whether the container can operate reliably throughout the year. Common options include dry coolers, cooling towers, chillers, plate heat exchangers, and hybrid cooling systems.

The best option depends on several project-specific factors:

Local annual temperature profile

Maximum summer ambient temperature

Humidity conditions

Water availability

Electricity cost

IT load profile

Operating hours

Maintenance capability

Future expansion requirements

A cooling solution that performs well in a cool climate may not be suitable for a high-temperature location. For this reason, thermal design should be based on local environmental data rather than copied from another project.

Power Distribution and Monitoring Systems

A high-density liquid-cooled container requires a reliable electrical system. Typical equipment includes UPS systems, PDUs, switchboards, busbars, ATS systems, grounding systems, and emergency power interfaces.

The monitoring system should cover:

IT power load

CDU supply and return temperatures

Flow rate and pressure

Pump status

Leak alarms

Power quality

Ambient temperature and humidity

Fire protection and access control

Energy consumption and PUE-related data

Remote alarm and maintenance records

Integration with a DCIM platform, BMS, or customized monitoring platform can improve visibility, reduce response time, and support preventive maintenance.

Cold Plate Liquid Cooling vs. Immersion Cooling

The two most common liquid cooling approaches for containerized data centers are cold plate liquid cooling and immersion cooling.

Cold Plate Liquid Cooling

Cold plate cooling uses liquid-cooled plates installed directly on major heat-generating components such as CPUs and GPUs. Heat is transferred from the component to the cold plate and then carried away by the cooling liquid.

Cold plate liquid cooling is widely used in AI servers, GPU clusters, HPC systems, and enterprise data center environments.

Its main advantages include:

Compatible with mainstream server architectures

Suitable for AI, GPU, and HPC workloads

Easier integration with rack-based data center infrastructure

Familiar maintenance processes for IT teams

High rack-density capability

However, cold plate cooling may not eliminate all airflow requirements. Components such as memory, power supplies, storage devices, and network cards may still require partial air cooling. The liquid cooling coverage ratio and residual air-cooling load should be defined during the design stage.

Immersion Cooling

Immersion cooling places servers or mining equipment directly into a dielectric cooling fluid. The fluid absorbs heat from the equipment and transfers it to an external heat exchanger.

Immersion cooling is often used in high-density mining operations, specialized computing equipment, high-temperature environments, and locations where dust control and noise reduction are important.

Its main advantages include:

High heat-transfer efficiency

Reduced fan power and noise

Strong resistance to dust and harsh environments

Suitable for high-power mining hardware

Reduced dependence on traditional air-conditioning systems

However, immersion cooling requires careful consideration of equipment compatibility, coolant selection, maintenance procedures, service workflow, and supply-chain support.

For mainstream AI servers and standard enterprise IT equipment, cold plate liquid cooling is often easier to adopt. For mining hardware and specialized high-heat-density systems, immersion cooling may provide stronger environmental and operational advantages.

Typical Applications

AI Data Centers and GPU Clusters

AI training and inference workloads require large GPU clusters with high power density. A liquid-cooled container data center can support rapid deployment of GPU infrastructure for cloud service providers, AI computing platforms, research institutions, and private enterprise AI projects.

HPC and Scientific Computing

HPC systems are used for weather modeling, engineering simulation, life sciences, material research, financial analysis, and scientific computing. These workloads often require dense compute clusters and stable thermal performance, making liquid-cooled modular data centers a practical option.

Edge Computing and Modular Data Centers

Manufacturing parks, telecom sites, energy facilities, ports, transportation hubs, and remote industrial locations may not have the time or space to build a conventional data center. Containerized liquid cooling provides a modular approach that can be deployed in phases as computing demand grows.

Hydro Mining and High-Density Computing

For hydro miners and other high-power computing equipment, liquid-cooled containers can reduce operating temperature, fan noise, dust exposure, and thermal stress. They are especially valuable in regions with available power resources, challenging environmental conditions, or fast deployment requirements.

Deployment Process for a Liquid-Cooled Container Data Center

1. Requirement Assessment and System Design

The project should begin with a clear definition of IT load, server type, rack quantity, rack power density, deployment location, electrical capacity, cooling conditions, network access, and expansion plans.

Key questions include:

What is the total IT load?

What is the target rack power density?

Will the project use cold plate or immersion cooling?

Is N+1 or 2N redundancy required?

What cooling source is available on site?

Is year-round free cooling feasible?

Are there transport, lifting, site-access, or foundation-load limitations?

Will the project require phased expansion?

2. Factory Integration and Testing

One of the main advantages of a containerized data center is factory prefabrication. Racks, CDUs, piping, power systems, monitoring systems, fire protection, and control systems can be installed and tested before shipment.

Recommended factory testing includes:

Hydrostatic pressure testing

Pressure holding tests

Pipe flushing and cleaning

Leak detection tests

CDU flow and pressure testing

Electrical system testing

Monitoring and control integration testing

Simulated load testing

Factory Acceptance Testing (FAT)

3. Site Installation and Commissioning

After delivery, the container must be positioned and connected to the site infrastructure. Typical work includes foundation placement, lifting, power connection, cooling-loop connection, network connection, and final commissioning.

Key commissioning tasks include:

Cooling-loop flushing

Supply and return verification

Valve and pressure adjustment

CDU parameter calibration

Leak-alarm testing

IT equipment installation and connection

Full-load or simulated-load testing

Remote monitoring platform integration

4. Operations and Continuous Optimization

After commissioning, the system should follow a structured maintenance plan. This may include coolant testing, filter maintenance, pump inspection, quick-coupling inspection, leak sensor testing, and energy-performance analysis.

Long-term operational reviews should evaluate:

PUE trend

Supply-return temperature difference

Pump energy consumption

Heat-rejection efficiency

Server failure rate

Coolant condition

System redundancy performance

Hydraulic balance after expansion

Five Common Design Mistakes to Avoid

1. Focusing Only on Cooling Capacity

A CDU may have enough nominal cooling capacity, but the full system can still underperform if flow rate, pressure drop, manifold sizing, pipe length, or server interface requirements are not properly matched.

Liquid cooling should be designed as a complete thermal management system, not as a single equipment purchase.

2. Ignoring Local Climate Conditions

A dry cooler solution that works well in a cool region may not maintain stable cooling performance in a hot climate. Cooling source selection should be based on local temperature, humidity, water availability, and annual operating conditions.

3. Neglecting Maintenance Access

Container space is limited. If maintenance aisles, valve access, filter replacement space, and server service clearance are not considered during design, operational costs can increase significantly.

High-density design should never eliminate practical serviceability.

4. Underestimating Leak Management

Liquid cooling systems can operate reliably, but leak prevention and detection must be included from the beginning. A complete strategy may include high-quality quick couplings, leak sensors, zoning alarms, automatic shutoff logic, pressure monitoring, and preventive inspection.

5. Failing to Reserve Expansion Capacity

AI server power density continues to increase. Without reserved capacity in power distribution, cooling equipment, pipework, network infrastructure, and site layout, future expansion may require expensive redesign.

How to Evaluate a Liquid-Cooled Container Data Center Supplier

When selecting a supplier, price should not be the only decision factor. The supplier should demonstrate system integration capability, engineering experience, testing procedures, and long-term service support.

Key evaluation points include:

In-house liquid cooling design and manufacturing capability

Ability to provide CDUs, liquid-cooled racks, cooling equipment, and container integration

Experience with cold plate and immersion cooling projects

Customization capability for rack size, power density, and interfaces

Factory testing and on-site commissioning capability

Remote monitoring and after-sales service

International shipping, installation, and technical support capability

Quality control procedures and relevant certifications

For large projects, request technical documents such as system diagrams, P&ID drawings, thermal-load calculations, flow calculations, equipment lists, test reports, and project references.

All-Liquid-Cooled Containerized Data Center Solution:

Lianli Liquid Cooling offers both 40-foot and 20-foot containerized data center options, with full customization available to accommodate server lineups from NVIDIA, AMD, Huawei, and others. As the direct manufacturer, we provide a comprehensive one-stop service—covering site assessment and planning, customization and integration, installation and commissioning, as well as maintenance and support. We offer rapid delivery (as fast as 45 days) and achieve a PUE as low as 1.15. Our products hold authoritative international certifications such as RoHS, CE, and UL, earning widespread trust for their technical excellence and service quality.

Conclusion

Liquid-cooled container data centers are becoming an important infrastructure option for high-density computing. They are not simply a replacement for traditional data centers; they are a modular deployment model designed for faster delivery, higher rack density, improved thermal management, and scalable expansion.

As AI platforms, GPU clusters, HPC systems, and high-power computing equipment continue to evolve, liquid cooling will become increasingly important. The most successful projects are built on complete planning across thermal design, cooling architecture, electrical distribution, factory integration, site commissioning, and long-term maintenance.

For organizations planning an AI data center, HPC cluster, edge computing site, or hydro mining project, a properly designed liquid-cooled container data center can provide a practical path toward faster deployment and more efficient high-density computing.

FAQ：

1.What size project is suitable for a liquid-cooled container data center?

Liquid-cooled container data centers can support projects ranging from several hundred kilowatts to multiple megawatts. The final capacity depends on server type, rack density, cooling architecture, external heat-rejection equipment, and site electrical conditions.

For AI GPU clusters, HPC systems, cloud nodes, and hydro mining projects, rack loads may reach 30 kW, 50 kW, 80 kW, or more. Compared with conventional air-cooled facilities, liquid-cooled containers can support higher IT load in a smaller deployment footprint and can be expanded through additional modules.

2.Which is better for a liquid-cooled container data center: cold plate cooling or immersion cooling?

Cold plate cooling is generally better suited for AI servers, GPU servers, HPC clusters, and standard data center equipment. It is compatible with mainstream server architectures and supports familiar maintenance processes.

Immersion cooling is often more suitable for hydro mining, specialized computing hardware, high-temperature environments, and applications with demanding dust or noise-control requirements. The final choice should consider equipment compatibility, power density, maintenance workflow, coolant type, and project budget.

3.What site conditions are required before deployment?

Before deployment, the project team should confirm foundation load capacity, transport and lifting access, electrical capacity, network connectivity, external cooling interfaces, drainage conditions, and future expansion space.

For cold plate liquid cooling systems, the design should also confirm the primary cooling source, whether dry coolers, cooling towers, chillers, or hybrid systems are required, as well as supply temperature, return temperature, flow rate, and pressure requirements.

Recommended Products：

20FT Liquid-Cooled Containerized Data Center | Modular/Prefabricated Containerized Data Center

40FT Liquid-Cooled Containerized Data Center | Modular/Prefabricated Data Center | Nvidia GB300 NVL72