Understanding the real environmental footprint of modern data centers through engineering, not assumptions.
Data centers are becoming an increasingly visible part of Indonesia’s infrastructure landscape.
As AI, cloud computing, digital services, and data-intensive applications continue to grow, so does public attention toward the infrastructure behind them.
And recently, two questions have become particularly prominent:
How much water does a data center actually use?
And:
How much noise does a data center really generate?
These are legitimate questions.
But they are also questions that require context.
A data center does consume energy. Some data centers consume significant amounts of water. Mechanical and electrical systems can generate noise.
But the actual environmental impact depends heavily on how the facility is designed, what cooling technology it uses, how large it is, where it is located, and how the infrastructure is operated.
That distinction matters.
Because the discussion should not be about automatically defending or criticizing data centers.
It should be about understanding the numbers behind the infrastructure.
The Question Is Not Simply: “Do Data Centers Use Water?”
They do.
The more useful question is:
How much water does a particular data center use, and what determines that number?
Water consumption in a data center is closely connected to its cooling architecture.
The U.S. Department of Energy (DOE) explains that for data centers using cooling towers, water consumption is related to the heat load generated by IT equipment and other facility loads, as well as the efficiency of the heat-removal system. DOE uses Water Usage Effectiveness (WUE) to measure data-center water performance.
The metric is:
WUE = Annual Site Water Usage ÷ IT Equipment Energy
expressed in liters per kilowatt-hour (L/kWh).
This is important because two data centers with similar IT capacity can have very different water requirements.
The cooling technology can change the equation significantly.
Not Every Data Center Uses the Same Amount of Water
There is no single number that can accurately describe the water consumption of every data center.
For example, evaporative cooling and cooling towers consume water as part of the heat-rejection process.
DOE explains that cooling towers primarily consume water through evaporation, with additional losses occurring through blowdown, leaks, and drift.
On the other hand, facilities using air-cooled systems or closed-loop cooling architectures can substantially reduce their direct water requirements.
The DOE has specifically noted that modern closed-loop cooling systems can recirculate water through sealed systems and heat exchangers rather than continuously consuming water through evaporation.
This means that asking:
“How much water does a data center use?”
without asking:
“What cooling system does it use?”
does not provide the complete picture.
So, How Much Water Are We Actually Talking About?
This is where numbers become useful—but only when their context is preserved.
The Jakarta Post, citing Hendra Suryakusuma, Chairman of the Indonesia Data Center Provider Organization (IDPRO), reported that water consumption can vary significantly according to cooling technology.
The report cited a global benchmark in which hyperscale data centers could consume approximately 1–5 million liters of water per day, particularly where evaporative cooling systems are used. It also noted that air-cooled chillers can significantly reduce water dependency.
That is a very large range.
And that is precisely the point.
1 million liters per day and 5 million liters per day are not the same infrastructure scenario.
Neither number should automatically be applied to every data center in Indonesia.
The actual figure depends on:
- IT load
- Cooling architecture
- Local climate
- Operating temperature
- Humidity
- Cooling efficiency
- Water-side system design
- Economizer strategy
- Facility utilization
DOE similarly notes that water consumption is influenced by heat load, cooling efficiency, climate, temperature and humidity control, and cooling strategy.
A Current Indonesian Example Shows Why Context Matters
The current public discussion surrounding the planned data center development around Jatiluhur, West Java, provides a useful illustration.
Recent reporting stated that the project’s future development could require approximately 384 liters of water per second, equivalent to roughly:
33.2 million liters per day
if that flow were maintained continuously for 24 hours.
However, the reported figure represents a future development stage, rather than an indication that the facility is already consuming that amount.
The same report stated that the initial requirement was substantially lower and would increase as the project expanded.
This distinction is important.
A projected maximum requirement is not the same thing as:
- current consumption,
- average consumption,
- peak consumption,
- or water actually withdrawn from a particular source.
It is therefore important for environmental discussions to distinguish between design capacity, projected demand, actual consumption, and source availability.
Water Consumption Should Be Viewed Alongside Water Availability
There is another important dimension.
A number by itself does not tell us whether water use represents a significant local impact.
For example:
1 million liters per day can mean something very different depending on whether the facility is located in an area with abundant industrial water infrastructure or an area already experiencing water stress.
This is why water assessment should consider:
Volume
How much water is required?
Source
Where does the water come from?
Availability
Can the local water system sustainably provide it?
Competition
Does the facility compete with agriculture, households, industry, or other users?
Technology
Can the cooling design reduce water dependency?
Lifecycle
How will water demand change as the facility expands?
This is particularly relevant in Indonesia.
IDPRO has recently emphasized that concerns regarding water, electricity, carbon emissions, and community impacts are legitimate considerations, but that they should be assessed proportionally and based on the actual technology and data used by each facility.
That is an important principle for the broader discussion.
What About the Noise?
The second issue receiving increasing attention is noise.
Here again, the first question should be:
What exactly is making the noise?
A data center is not simply a room full of computers producing one uniform level of sound.
The major sources of external mechanical and electrical noise can include:
- Cooling towers
- Air-cooled chillers
- Air-handling equipment
- Pumps
- Generators
- Transformers
- Other mechanical plant equipment
ASHRAE specifically identifies heat-rejection equipment such as cooling towers and air-cooled chillers, as well as emergency and prime power-generating equipment, as important sources of data-center sound emissions.
Recent analysis from UNSW similarly explains that the characteristic low-frequency hum associated with data centers generally comes from cooling and mechanical infrastructure, including air-handling units, chillers, cooling towers, pumps, and backup generators.
So saying:
“The servers are making the neighborhood noisy.”
is technically too simplistic.
The more accurate description is:
The mechanical and electrical infrastructure required to operate and cool the data center can generate noise.
But Equipment Noise Is Not the Same as Community Noise
This may be one of the most important distinctions in the current discussion.
A piece of equipment can produce a relatively high sound level at its source.
That does not automatically mean residents outside the facility experience the same sound level.
Between the equipment and the surrounding community, there can be:
Equipment → Building → Acoustic Treatment → Barriers → Distance → Property Boundary → Community
Each element can affect the resulting sound level.
ASHRAE notes that community sound levels from exterior heat-rejection and power-generation equipment are typically assessed against applicable local or regional noise requirements, with limits often referenced at property lines or locations where a potential complainant may be affected.
UNSW similarly notes that noise from equipment inside or around data centers can be significantly reduced through distance, barriers, acoustic treatment, equipment selection, and building design before reaching surrounding areas.
Therefore:
85 dBA at the equipment
does not mean:
85 dBA at the nearest house.
Those are different measurement points and different questions.
Does That Mean Data Center Noise Is Not a Problem?
Not necessarily.
The correct conclusion is more nuanced.
Data center mechanical systems can produce significant noise, particularly cooling equipment and backup power systems.
ASHRAE recognizes this as an engineering consideration and recommends that sound from exterior equipment be addressed through appropriate mitigation and compliance with applicable community noise requirements.
The question is therefore not:
“Are data centers noisy or not?”
It is:
“What is the sound level at the relevant receptor, under what operating condition, and how effectively has the facility been designed to control it?”
That is a much more meaningful engineering question.
Noise Also Depends on Operating Conditions
A data center does not necessarily operate all equipment at maximum output all the time.
Noise can vary depending on:
- IT load
- Ambient temperature
- Cooling demand
- Number of cooling units operating
- Fan speed
- Chiller operation
- Generator testing
- Emergency operation
- Equipment configuration
For example, a cooling system may increase its mechanical activity during periods of higher heat load.
Backup generators may normally remain off but operate during testing or power interruptions.
UNSW notes that backup generators can operate during power disruptions and may also be run for maintenance testing.
This means that a proper noise assessment should consider normal, peak, test, and emergency operating scenarios, rather than relying on one measurement taken under one condition.
ASHRAE explicitly recommends considering exterior equipment during normal, emergency, and test operation.
The Real Environmental Footprint Is a Design Question
This brings us to a broader point.
The environmental footprint of a data center is not determined simply by the fact that it is a data center.
It is influenced by engineering decisions.
Cooling
What cooling architecture is used?
Water
How much water is consumed, and where does it come from?
Energy
How efficiently is electricity converted into useful IT output?
Noise
Where are mechanical systems located, and how is sound controlled?
Operations
How are temperature, humidity, equipment utilization, and maintenance managed?
Expansion
How will resource requirements change as the facility scales?
This is why infrastructure planning matters.
The Question Should Move From “How Big Is the Impact?” to “How Is It Designed?”
A more useful way to evaluate a data center is through a series of measurable questions.
| Issue | Question to Ask |
| Water | How many liters per day does the facility require under actual operating conditions? |
| Cooling | What cooling technology is being used? |
| WUE | What is the facility’s Water Usage Effectiveness? |
| Water Source | Where does the water come from? |
| Availability | What is the condition of the local water supply? |
| Noise | What are the major noise sources? |
| Sound Level | What is measured at the relevant property boundary or receptor? |
| Operating Condition | Is the measurement during normal, peak, test, or emergency operation? |
| Mitigation | What acoustic and engineering controls are installed? |
| Lifecycle | How will resource requirements change as capacity increases? |
This shifts the conversation from headline numbers to measurable infrastructure performance.
Data Centers Should Be Measured, Not Assumed
The current debate around data centers is understandable.
They are large infrastructure assets.
They require significant electrical capacity.
They generate heat.
Some cooling architectures consume substantial amounts of water.
And their mechanical and electrical systems can generate noise.
Those impacts should not be dismissed.
But neither should they be generalized.
A 1 MW facility and a 100 MW facility are not equivalent.
An evaporative cooling system and a closed-loop system are not equivalent.
A cooling tower located close to a residential boundary and one separated by significant distance are not equivalent.
A projected water requirement and actual consumption are not equivalent.
And equipment noise measured at the source is not equivalent to community noise measured at the property boundary.
The engineering details matter.
A More Balanced Way to Talk About Data Center Growth
As Indonesia continues to expand its digital infrastructure, the conversation should evolve.
Instead of asking only:
“How much water does a data center use?”
we should ask:
“How much water does this facility use, why does it use that amount, and can the design reduce it?”
Instead of:
“How noisy is a data center?”
we should ask:
“What equipment generates the noise, what are the measured levels at the relevant receptors, and what controls are in place?”
And instead of:
“Are data centers good or bad for the environment?”
we should ask:
“How effectively is the infrastructure designed, operated, measured, and managed?”
The DataGarda Perspective: Infrastructure Requires Context
At DataGarda, we believe infrastructure decisions should be evaluated through data, engineering context, operational performance, and lifecycle thinking.
The objective is not to argue that data centers have no environmental impact.
Nor is it to assume that every data center creates the same level of impact.
The objective is to understand the infrastructure accurately.
Measure the water.
Measure the energy.
Measure the noise.
Understand the cooling architecture.
Understand the operating conditions.
Assess the local environment.
Then design and operate accordingly.
Because infrastructure readiness is not simply about building more capacity.
It is about ensuring that the infrastructure can scale responsibly, operate efficiently, and remain resilient within the environment where it exists.
Conclusion: Beyond the Headline
The question is not whether data centers consume water.
They do.
The question is not whether data centers can generate noise.
They can.
The more important question is:
How much—and under what conditions?
Water consumption depends on cooling technology, IT load, climate, and operational strategy.
Noise depends on mechanical and electrical equipment, operating conditions, distance, acoustic treatment, and the point at which sound is measured.
That is why the environmental discussion around data centers should move beyond generalized numbers.
Measure the impact. Understand the engineering. Design the infrastructure.
Because better infrastructure decisions begin with better questions.
DataGarda Call to Action
Planning, expanding, or assessing a data center?
Environmental performance should be considered alongside infrastructure readiness, operational efficiency, resilience, and lifecycle requirements.
Talk to DataGarda about building data center infrastructure that is designed not only to scale—but to operate responsibly and with confidence.
Build for Growth. Operate With Control.
References & Sources
- U.S. Department of Energy — Cooling Water Efficiency Opportunities for Federal Data Centers
Explains WUE, cooling-tower water consumption, evaporative losses, and how cooling strategy and operating conditions influence water use. - U.S. Department of Energy — The Truth About AI Data Centers
Discusses closed-loop cooling and how recirculating cooling systems can substantially reduce direct water consumption. - U.S. Department of Energy — Estimating Methods for Determining End-Use Water Consumption
Provides technical explanation of evaporation, blowdown, drift and other sources of cooling-tower water consumption. - ASHRAE Handbook — Data Centers and Telecommunication Facilities
Covers data-center noise sources, including cooling towers, air-cooled chillers, generators and other mechanical/electrical equipment, as well as community noise assessment at property boundaries. - The Jakarta Post — “Indonesia’s data center boom raises alarm over water use”
Provides Indonesia-specific context on data-center cooling technologies and cites a benchmark of approximately 1–5 million liters/day for some hyperscale facilities using evaporative cooling. - Katadata — “IDPRO: Pembangunan Data Center Harus Memperhatikan Aspek Lingkungan”
Reports IDPRO’s position that water, energy, carbon emissions and community impacts should be considered, while emphasizing proportional, data-based assessment according to the technology actually used by each facility. - Bisnis.com — Jatiluhur data-center water requirement
Reports that the planned development could eventually require approximately 384 L/s, equivalent to roughly 33.2 million liters/day at continuous flow, while noting that the requirement would increase progressively rather than occurring at that level from the initial stage. - UNSW — “Are data centres actually noisy, and is this bad for our health?”
Explains the main sources of data-center noise and the role of distance, barriers, acoustic treatment and equipment selection in reducing noise reaching surrounding areas.
