When a data center developer says the data center will use a closed cooling system, it usually means water use will be much lower than in traditional evaporative cooling, but it does not automatically mean the facility will use no water at all.
What “closed cooling system” usually means
In data center language, a closed cooling system (often called closed-loop cooling) generally refers to a setup where a coolant – water, water-glycol mix, or another liquid – recirculates in sealed pipes and heat exchangers instead of being sprayed into the air and lost through evaporation. In contrast, open-loop evaporative systems rely on cooling towers that evaporate large amounts of water to reject heat, which is why they can consume millions of gallons per year at scale. A properly designed closed-loop system can dramatically reduce on-site water consumption because the same fluid is reused rather than continuously replaced. However, “closed” describes the circuit that carries heat away from the servers, not the entire water footprint of the data center. Even if the primary loop is sealed, there may still be make-up water to replace small losses, domestic water for toilets and sinks, humidification in some climates, and indirect water use associated with electricity generation.
Closed loop vs water-free cooling
It’s important to distinguish between “closed cooling” and truly water-free cooling. Closed-loop systems can be:
- Liquid-based, using water or a water mixture in a sealed circuit.
- Refrigerant-based, using mechanical chillers and refrigerant rather than process water.
- Hybrid systems, combining air-cooled equipment with a small closed water loop.
Only solutions specifically described as “water-free” or “air-cooled” can reasonably be expected to operate with negligible process water use. Even then, the facility still has non-cooling water needs and an indirect water footprint through its power supply.
How much water is still used?
In practice, closed-loop water systems often cut cooling water consumption by orders of magnitude compared with evaporative cooling, especially for large, high-density sites. Many hyperscale and AI-focused facilities are moving in this direction exactly to reduce strain on local water supplies and to meet regulatory or community expectations. Some designs use adiabatic coolers sparingly or only above certain temperatures, which means there can still be occasional water use for peak conditions. Others rely on chillers and dry coolers year-round, trading higher electricity use for lower water demands. Because each project is different, the only reliable way to know if a specific data center will “not consume water” for cooling is to ask for details: the water source, expected annual water use, and whether any evaporative or adiabatic equipment is planned. Claims that a closed system uses “no water” are sometimes shorthand in public discussions and can gloss over small but real volumes.
Questions to ask a developer
If you’re evaluating a proposal or trying to understand impacts, questions like these are useful:
- Is any evaporative or adiabatic cooling equipment included, even as backup or for peak days?
- What is the expected annual cooling water use, and how does that compare to local municipal or industrial demand?
- Is the cooling loop water-based, refrigerant-based, or a hybrid?
- How will blowdown, leaks, and maintenance be handled, and is there a make-up water allowance?
- What are the non-cooling water uses on site (sanitary, landscaping, humidification)?
Taken together, these answers will tell you whether “closed cooling system” in this particular project merely reduces water use significantly or truly approaches near-zero water consumption for cooling, rather than leaving you to rely on the label alone.