
As data centers face rising thermal loads driven by AI, HPC, and increasing rack densities, traditional air cooling methods are hitting their limits—especially in environments where space and efficiency matter.
Liquid cooling can solve this, but it’s not plug-and-play. There are several methods, each with different requirements and trade-offs. The key is knowing what fits your environment—and how to roll it out without overhauling everything.
At Chatsworth Products, we’ve helped organizations deploy liquid cooling infrastructure in practical, scalable ways—getting more performance per watt without the need for a full rebuild.
This guide covers the most common liquid cooling types, when hybrid strategies make sense, and how to plan your next steps with confidence.
Common Liquid Cooling Types (and When to Use Them)
Liquid transfers heat far more efficiently than air—but not all methods are alike. Here are the main types used in data centers, and when each makes sense.
Direct-to-Chip (D2C) Cooling
Direct-to-chip, also known as D2C or DTC, cooling brings the coolant right to where the heat is being generated—the chip itself. Liquid coolant flows through metal plates, called cold plates, that are mounted directly on top of CPUs or GPUs. As the chips heat up, the cold plates pull the heat into the liquid, which is then carried away, cooled, and recirculated.
- Single-phase D2C uses a liquid that stays in liquid form while it carries heat away from the chip.
- Two-phase D2C uses a liquid that boils into vapor when it touches the hot chip surface. That phase change—liquid to gas—absorbs a lot of heat very quickly, making it even more efficient.
Use Cases and Benefits:
Because this method pulls heat directly from the chip, it’s fast and efficient helping prevent overheating and keeping performance steady, even under heavy workloads.
- Ideal for AI and HPC environments where heat builds up quickly
- Great for data centers with limited space that need to maximize compute density
CPI’s , integrated with ZutaCore® HyperCool, uses a waterless, two-phase D2C cooling system to safely and efficiently cool high-performance equipment.
See how it works in this demo:
Rear Door Heat Exchangers (RDHx)
RDHx replaces the back door of a cabinet with a liquid-cooled coil. As hot air exits the servers, it passes through the coil, transferring heat to the liquid before the air enters the room.
The liquid stays contained inside the door—never touching the equipment. Instead, it cools the airflow coming out of the servers, reducing the need for large-scale air conditioning systems.
Use Cases and Benefits:
RDHx is a simple way to boost cooling capacity—especially in air-cooled environments—without modifying the IT hardware.
- Ideal for existing air-cooled setups with a few hotspots
- Commonly used as a drop-in retrofit—no need to overhaul the whole room
- Helps improve airflow containment at the rack level
- A smart step into liquid cooling without jumping straight to direct-to-chip systems
Think of RDHx as a middle-ground solution: it doesn’t cool the components directly but is a practical step toward liquid cooling without major changes.
Chilled Water Systems
Chilled water cooling works a bit like an air conditioner for your data center. Cold water is piped through coils or cooling panels, and as warm air flows over them, the heat transfers into the water. That heated water is then pumped out, cooled by a chiller, and recirculated.
Picture rows of servers with chilled coils nearby—either under a raised floor, above the rack, or in between cabinets—quietly pulling heat from the air as it moves.
Use Cases and Benefits:
Chilled water systems are common in large, centralized data centers where robust, building-wide cooling infrastructure already exists.
- Often tied into a facility’s HVAC or cooling tower system
- Frequently used with raised floors or in-row cooling setups
- A proven option for steady, room-level cooling across large footprints
While chilled water is stable and widely used, it cools the environment, not the IT components directly. As power densities increase, it may struggle to keep up on its own.
Immersion Cooling
This is what most people picture when they think of liquid cooling: entire servers fully submerged in a tank filled with a special, non-conductive liquid. As the components generate heat, the surrounding fluid absorbs it directly. That heated liquid is then cooled and circulated back into the tank.
Use Cases and Benefits:
Immersion cooling is typically reserved for extreme use cases where traditional airflow can’t keep up.
- Used in very high-density or edge environments with limited airflow options
- Common in crypto mining operations and advanced HPC research labs
- Rare in traditional enterprise due to hardware compatibility and maintenance complexity
Immersion offers top-tier thermal performance but comes with challenges—maintenance, hardware handling, and compatibility issues—making it less common in traditional enterprise setups.
Type of Liquid Cooling | When to Use It | Why It Works |
|---|---|---|
D2C (Two-Phase) | Supporting AI/HPC workloads with 20kW+ per rack | Maximizes heat removal directly at the chip; waterless, scalable retrofits |
RDHx | Adding cooling to existing racks without full system changes | Efficient air-liquid hybrid for gradual upgrades; no need to change server internals |
Chilled Water | Large data halls with existing building cooling systems | Centralized cooling via facility HVAC; good for consistent, high-volume loads |
Immersion | Highly dense deployments with limited airflow options | Total liquid submersion offers max thermal transfer |
Integrating Liquid Cooling into Existing Infrastructure: A Hybrid Approach
Most data centers weren’t built for today’s high-density compute loads—but shutting down for a full rebuild isn’t an option.
That’s why many are turning to hybrid cooling—a strategy that blends traditional air cooling with targeted liquid cooling to boost efficiency, manage heat, and scale up faster.
What hybrid cooling looks like:
- Liquid cooling at the rack or component level, typically focused on high-power CPUs and GPUs
- Air cooling supporting lower-density equipment and the broader environment
Direct-to-chip cooling is ideal for hybrid setups because it targets only the hottest components, integrates with standard racks, and works alongside existing cooling systems.
Real-World Example: CPI’s ZetaFrame® + ZutaCore® Solution
At the University of Chicago, CPI’s hybrid approach was successfully put to the test. CPI delivered the —a waterless, two-phase direct-to-chip solution, tested and ready to deploy without a full facility overhaul.
Why it works so well in hybrid environments:
- uses a non-toxic, non-conductive coolant that evaporates safely in the unlikely event of a leak—protecting both hardware and the environment. It’s maintenance-free, doesn’t require fluid replacement, and has an ultra-low global warming potential (GWP).
- is designed for high-performance environments. It includes pre-installed airflow management accessories—perfect for hybrid deployments where efficiency and containment matter.
- With the highest load rating on the market (5,000 lb. static and 4,000 lb. dynamic), ZetaFrame is built to support larger cooling components typical of liquid-cooled, high-density racks.
And because CPI’s ZetaFrame Cabinet offers more than just cooling—including pre-installed PDUs, built-in cable management accessories, and more—it simplifies the complexity that often comes with adding liquid cooling.
What sets CPI apart:
- Faster, lower-risk liquid cooling deployment
- A scalable solution—start small, expand as needed
- Future-ready infrastructure, proven in real-world environments
- Expert guidance and support from a partner who understands liquid cooling
Three Critical Considerations Before You Deploy
Liquid cooling isn’t just about choosing a technology—it’s about making it work in the real world, with the infrastructure, constraints, and teams you already have.
Here are three things to get right before you deploy:
1. Get More from Your Cooling by Getting Airflow Right
Liquid cooling doesn’t replace airflow management—it relies on it. Before deploying any advanced cooling system, seal your airflow leaks. Bypass and recirculation can kill efficiency, no matter how advanced your cooling setup is.
And it’s not just about aisle containment. The biggest improvements often happen inside the cabinet:
- Use to cover unused rack spaces
- Add grommets and brush strips to seal around cable openings
- Prevent hot air recirculation within the rack with a especially in facilities without full hot aisle containment.
Airflow fixes are simple, low-cost, and highly effective. They boost efficiency and can even delay the need for liquid cooling—helping you get more out of your current setup before scaling further.
2. Equip Your Team with Guidance and Support
Liquid cooling takes more than hardware. It often requires specialized planning, installation, and maintenance. your team will need the right expertise to:
- Plan around thermal zones and rack-level configurations
- Handle liquid lines, fittings, and pressure systems
- Monitor and maintain new components safely over time
If your team hasn’t worked with liquid cooling before, consider external support. Experienced partners like CPI can help your team avoid costly missteps, streamline deployment, and set your team—and your infrastructure—up for long-term success.
3. Check Infrastructure Compatibility
Liquid cooling affects more than just the cooling system; its success depends on how well these systems integrate. Before deploying, take a close look at:
- Racks: Do they support the weight and dimensions of liquid-cooled gear?
- Power distribution: Can your PDUs handle higher densities and additional equipment?
- Cable management: Are your pathways organized and sealed to avoid blocking airflow or coolant lines?
- Monitoring tools: Do you have sensors or DCIM systems that can track thermal changes in real time?
- Physical layout: Is there space for cooling manifolds, return lines, or any added components?
Modular platforms like CPI’s ZetaFrame® Cabinet System support phased deployment, so you can evolve your infrastructure without starting from scratch.
Don’t Tackle Liquid Cooling Alone—Here’s How CPI Can Help
Liquid cooling is a system-wide change. That’s why you need more than components—you need a partner.
At Chatsworth Products (CPI), we design and manufacture innovative IT infrastructure that powers and protects critical technology worldwide. But we don’t stop at hardware—we bring deep expertise in thermal management and integrated cooling strategies to help you scale with confidence.
Whether you’re retrofitting for hybrid cooling or deploying high-density, liquid-cooled racks, CPI delivers the infrastructure, insight, and support to make it work in your environment.
or get in touch with our team to
