When a data center expands, added server capacity comes with a second question: where will the extra heat go on the hottest, most humid operating days? Data center cooling towers can form part of the answer where the site can support evaporative heat rejection. Their selection depends on the cooling load, the temperature the equipment can accept, and the water available to run the system.
In Southeast Asia, data center cooling projects are taking shape alongside long-term warming and growing competition for water and electricity. The International Energy Agency’s Energy and AI outlook expects the region’s data center electricity demand to more than double by 2030. That points to the growing scale of facilities. Each project still has to choose a heat rejection system that fits its location and equipment.

Illustration of a data center heat rejection plant.
For developers planning in 2027, the decisions that matter start well before procurement. Establish design weather and the cooling load by phase, identify who supplies and treats the water, and ask vendors to show how their equipment performs when a unit is out of service.
How data center cooling towers reject heat
A cooling tower transfers heat from recirculating water to outdoor air. This heat transfer takes place outside the data hall. In an evaporative cooling tower, a portion of the water evaporates and carries heat away. The cooler water returns to the cooling system. In a common chilled-water design, the tower serves the condenser water loop of a water-cooled chiller. The chilled water loop that serves the building remains separate.
That distinction matters when comparing data center cooling systems. Server-room air cooling uses air to move heat away from IT equipment. A computer room air handler may cool that air with chilled water, while a computer room air conditioner may use a direct-expansion refrigeration cycle. The chiller can then pass heat to condenser water, and the tower rejects it outdoors. Some air cooling designs use air-cooled equipment for heat rejection instead.
Liquid cooling takes heat from the IT equipment through a liquid circuit. Direct liquid cooling can use cold plates on selected components, while immersion cooling places compatible hardware in a dielectric fluid. A cooling distribution unit or heat exchanger may separate the IT loop from the facility loop. The heat still needs to reach the surrounding environment. A dry cooler, an evaporative cooling tower, or a hybrid system can do that, depending on the facility loop temperature and local conditions.
Higher rack densities and high-performance computing can push project teams to consider liquid cooling. The practical question is how much heat each system must remove, at what temperature, and under which operating conditions. Ask the IT and mechanical teams to agree on those figures before they select the cooling plant.
Air cooling and cold aisle containment
Many data centers use cold aisle containment to keep cold supply air separate from hot air leaving the racks. Containment still depends on the air conditioning system delivering cool air where the IT equipment needs it. Check rack densities and airflow before assuming additional fan speed will solve a hot spot. The heat generated by the servers must be carried to the cooling system regardless of how the aisles are arranged.
Liquid cooling and the facility heat exchanger
Data centers can use direct-to-chip liquid cooling while retaining air cooling for other components. Immersion cooling can simplify some air-side requirements, but it still needs pumps, heat exchange and controls. Confirm which cooling units remove heat from which loads. In a closed system, a plate heat exchanger can transfer heat between the facility water circuit and a separate IT coolant loop without mixing the fluids.
Some liquid cooling technology lets the facility operate with warmer supply water than a conventional chilled-water system. That can enable free cooling for more hours at a suitable site or make waste heat easier to recover if a compatible user is nearby. Ask for the temperature and flow required by the IT equipment before crediting either benefit. The combined heat from IT load, pumps and other relevant equipment still has to be reflected in the duty calculation, with equipment and electrical losses counted in the appropriate part of the system.
What a warming climate means for data center cooling
The World Meteorological Organization’s State of the Climate in Asia 2025 found that Asia warmed faster than the global land-and-ocean average from 1991 to 2025. The IEA’s Southeast Asia Energy Outlook 2026 projects nearly fivefold growth in regional electricity demand for space cooling by 2050 under its Stated Policies Scenario. Warmer temperatures contribute alongside rising air-conditioner ownership, income, and floor area. World Meteorological Organization climate findings.
These are regional findings. For evaporative cooling towers, the key weather input is the entering-air wet-bulb temperature, which reflects both temperature and moisture. The tower’s approach is the difference between leaving water temperature and entering-air wet-bulb temperature. At a higher wet-bulb temperature, it takes a closer approach to deliver the same leaving water temperature. Designers may need more tower capacity or a cooling system that can run with warmer water.
The cooling tower range is the difference between hot water entering and cooler water leaving the tower. Range depends on heat load and water flow. Approach describes how close the cooling tower can bring the water temperature to the entering air’s wet bulb. Keeping these two values separate helps buyers compare vendors at an equivalent duty point.
For a 2027 data center project, start with site-specific design weather and test cooling performance across operating conditions. Model full build-out and lower IT loads during early phases. Consider how adjacent equipment and site layout affect air intake. In a compact plant yard, warm, moist discharge air returning to an intake can raise the actual entering wet-bulb temperature. Check the layout as well as the equipment selection.
Compare cooling tower selections using the same heat load, water flow and entering-air conditions. Illustrative image.
Thailand’s draft 80% blowdown reuse rule
Thailand offers a practical example of why water needs its own design review. The Cabinet endorsed a broader data center policy framework on November 10, 2026. Separately, The Nation reported that a draft ministerial regulation scheduled for review on November 27 would require covered data centers to treat and reuse at least 80% of their monthly cooling tower wastewater discharge, known as blowdown, as makeup water. As of November 22, the 80% provision was still reported as a proposal. It concerns monthly tower blowdown volume. The Nation’s reporting on the draft rule.
Blowdown is water deliberately removed to control the concentration of dissolved solids. As evaporation removes water, minerals remain in the condenser loop. Blowdown helps limit their buildup. A water treatment programme also helps control scale, corrosion and biological growth. Makeup water replenishes water lost through evaporation, blowdown and drift. Recycling treated blowdown can reduce the need for fresh makeup water, while the tower still needs water to replace what evaporates.
Data center cooling water systems should be evaluated over ordinary and peak operating periods. A useful feasibility check starts with a monthly water balance. Record how much makeup comes in, how much blowdown leaves the tower, how much the treatment plant can return at the required quality, and how much residual discharge needs another route. Account for maintenance and downtime. Keep the water quality target tied to the cooling tower, chiller, pipework and local discharge requirements.
Treatment also takes space and power. Project teams should decide where meters, storage, sampling points and treatment equipment belong while the plant layout is still flexible. Make responsibilities explicit among the cooling system supplier, the treatment vendor and the site operator. The water balance needs to work on an ordinary operating month and during peak conditions.

Blowdown reuse involves more than the tower. Set aside space for treatment, metering and maintenance access. Illustrative image.
Compare evaporative cooling, dry cooling and hybrid systems
Data centers have several ways to reject heat. Evaporative cooling can achieve lower water temperatures than dry cooling under suitable conditions, but it requires makeup water and water treatment. Dry cooling uses outdoor air without routinely evaporating cooling water at the heat rejection equipment. Its capacity depends on dry-bulb temperature and the temperature difference available at the air-side heat exchanger. Hot outdoor conditions can make a dry system larger or increase fan power for a given duty.
Hybrid systems can use dry cooling when outside air conditions permit and add evaporative assistance when needed. Compare when the wet mode operates, how often water is available, and what happens if a unit fails during the most demanding hours. Dry cooling may be a better fit when water is constrained and the liquid cooling system can tolerate warmer water. A water-cooled chiller and tower may suit a different temperature requirement.
Even within a single cooling plant, conditions vary by hour and by phase of construction. Where climate and loop temperatures permit, a water-side economizer can use a cooling tower and a heat exchanger to cool the chilled water loop with reduced compressor energy. This form of free cooling depends on the available water temperature and equipment arrangement. Include pumps and fans in the calculation. Ask vendors to identify when free cooling is available at the actual site.
Open and closed-circuit cooling towers
In an open-circuit cooling tower, the circulating water contacts outside air directly. Fill media increases the surface area for heat and mass transfer. In a closed-circuit tower, the process fluid stays inside a coil, while a separate spray-water circuit and airflow carry heat away. A closed process circuit can protect the process fluid from airborne contamination, while the separate spray water still needs treatment and maintenance when the tower runs wet.
Compare energy efficiency, water use and operating costs
Ask suppliers of data center cooling systems for annual estimates at the same IT load profile and weather assumptions. Power usage effectiveness (PUE) compares total facility energy with IT equipment energy. Water usage effectiveness (WUE) relates water used to IT equipment energy. Specify the boundary for water usage so everyone knows which water sources and operations the number covers. These facility-level measures put the cooling plant’s energy and water trade-offs in context.
The IEA’s Energy and AI report estimates that data center cooling can account for about 7% of total electricity consumption in efficient hyperscale facilities and more than 30% in some less-efficient enterprise data centers. This spread shows why a single percentage gives a poor basis for design. A lower PUE estimate can come with higher water consumption when a design relies more on evaporation. A lower WUE estimate can come with higher fan or compressor energy. A whole-plant comparison should also include water availability during dry periods, discharge restrictions, treatment and chemical costs, electrical capacity, and planned maintenance. Use the same assumptions for all cooling solutions so owners can see which costs actually change.
For an existing facility, start with measured fan power, pump power, condenser water temperatures, makeup and blowdown volumes. A change in heat load may alter cooling energy and water usage. Compare readings collected under similar operating conditions and identify what can be addressed through controls, maintenance, or a thermal upgrade before specifying replacement equipment.
Plan commissioning and expansion before ordering
Phase the cooling infrastructure around the IT load that will actually come online. Modern data centers often expand in stages. Keeping all fans and pumps online for a partly occupied data hall can raise energy consumption. Ask how the control sequence stages tower cells, efficient chillers and pumps as loads change. Ask how it will prevent short cycling and maintain equipment flow limits when demand is low.
Run the same check at the other end of the load range. On a hot, humid design day, establish how much cooling capacity remains when a tower cell or pump is unavailable. Confirm whether the remaining cooling units can dissipate heat at that duty, and identify any load reduction the operator would have to make. If outside air conditions prevent an economizer from working, the cooling technology in reserve must still meet the agreed load.
Assign someone to verify the plant as a continuous process from IT load to outdoor heat rejection. Record the operating weather, water flowing through each circuit, leaving temperatures and equipment energy consumption during commissioning. Repeat the comparison after major expansion. That gives future operators a useful baseline for spotting loss of cooling efficiency, rising maintenance costs, or unexpected increases in fan and pump energy.
Tim Bozic’s 2027 outlook for commercial cooling towers
Tim Bozic of ICS Thailand estimates that demand for commercial cooling tower projects in Asia-Pacific could increase by roughly 50% in 2027. It is his personal forecast, based on 20 years of experience in the cooling tower industry. For buyers, the practical question is how quickly a rise in project activity could affect the time available to verify site conditions and compare cooling systems before ordering.
What to specify before requesting a cooling tower quote
- Design duty. Provide the heat rejection load by phase, circulation flow, target leaving water temperature and site design entering-air wet-bulb temperature. Ask vendors to identify which conditions control cooling capacity.
- IT and plant interfaces. Confirm the IT load and cooling requirements, the heat exchanger where loops meet, and the facility supply and return temperatures. State whether the cooling system will serve air cooling, liquid cooling, or both.
- Water and discharge. Provide makeup water quality and availability, discharge restrictions, and the treatment assumptions. Define how water usage, blowdown, and any reuse would be metered.
- Redundancy and access. State which tower cells, cooling units, pumps or chillers can be taken offline, plus the required output during maintenance. Allow space to service equipment without blocking other operating units.
- Performance evidence. Request fan and pump input, expected water use, and cooling performance at full and part load. Agree how performance will be verified during commissioning and what operating conditions will be recorded.
Bring the site’s design duty, water limits, operating hours and phasing to the first engineering discussion. Those inputs make it easier to compare what each system can deliver through a full season of operation.
Leave A Comment