Few pieces of infrastructure have gone from invisible to controversial as quickly as the data center. For years, they quietly powered our banking apps and e-commerce platforms. Today, they draw street protests. In Batam, residents around the Nongsa Special Economic Zone have repeatedly demonstrated after clean-water supplies to their homes were cut while, they say, water kept flowing to data centers inside the zone.

Their anger is understandable and deserves a serious response. But the national conversation risks sliding into one of two unhelpful positions: rejecting data centers outright or waving them through with minimal scrutiny. Neither serves Indonesia.

Start with a basic fact: data centers are not new here. Many have operated for years without much public attention. What has changed is the scale of computing demand, driven by generative AI, cloud services, and large-scale digital applications. Indonesia's installed data center IT capacity stood at roughly 580 megawatts in the first half of 2026, and some analysts project it could reach 3.5 gigawatts by 2030. Globally, the International Energy Agency expects data center electricity use to more than double between 2024 and 2030, to around 945 terawatt-hours, with AI-focused facilities growing even faster.

More computing means more heat, and heat must be removed. That is where water enters the story. Yet the point most often lost in the debate is this: not all data centers cool their servers the same way, and their water footprints can differ enormously.

Smaller facilities can rely on air-based cooling, such as air-cooled chillers, which need very little water, or none at all for the primary cooling process. Larger ones have a range of options, such as open-loop evaporative cooling towers, closed-loop systems, chilled-water systems, and, increasingly, direct liquid cooling that removes heat at the chip itself. Open-loop evaporative systems are energy-efficient but consume water continuously, because evaporation is how they shed heat. Closed-loop systems recirculate the same water through sealed pipes and need only small top-ups.

Vantage, a US data center operator, reports that its closed-loop campus in Wisconsin will use about 83,000 liters of water a day at peak, compared with roughly 19 million liters for a similar-sized campus using evaporative cooling. There are trade-offs, however. Low-water designs often consume more electricity, which matters in a country whose grid still relies heavily on coal. Choosing a cooling system is a balancing act between water, energy, cost, and local climate, not a moral choice between "good" and "bad" technology.

This is why the question "Do data centers use a lot of water?" is too blunt to be useful. The better questions are specific. What cooling technology will this facility use? How much water and electricity will it need, on average and at peak? Where will the water come from: municipal supply, groundwater, recycled wastewater, or desalination? What is the hydrological condition of the site? And how will the project affect the people who live next to it?

Location matters enormously. Batam, for instance, has no major rivers or groundwater aquifers and depends entirely on rainfall for its freshwater, according to BMI research. Its water utility has acknowledged that current infrastructure cannot simultaneously meet residential and data center demand as the market grows. A cooling design that is perfectly reasonable in one region may be irresponsible in another.

Water is not the only concern. Data centers draw large, constant loads from the grid, requiring careful power planning. Their cooling fans, chillers, and backup diesel generators can produce a persistent low-frequency hum; residents living near data centers in the United States and Europe have complained, and in some cases gone to court, over precisely that. Acoustic enclosures and sensible siting can address much of this, but only if noise is assessed before construction, not after complaints pile up.

All of this points to one policy conclusion: every data center project should be preceded by a transparent, site-specific feasibility study. It should cover energy and water requirements, the cooling technology chosen and why, the water source and its effect on local supply, grid capacity, noise, wider environmental impacts, and risks to surrounding communities. Its key findings should be published in plain language.

Transparency is currently the weakest link. Researchers note that reliable data on data center water use in Indonesia remains scarce, partly because companies treat it as proprietary information. Requiring operators to regularly report water usage effectiveness (WUE) and power usage effectiveness (PUE) to regulators and the public would be a simple, powerful start and would reward those already investing in efficient designs.

Transparency is not a burden on investment; it is a precondition for a social license to operate. Communities that understand what is being built and see that their basic needs come first, are far less likely to resist. Those kept in the dark while their taps run dry will push back, and rightly so.

Indonesia cannot afford to sit out the digital and AI economy. Data localization and the region's search for capacity beyond Singapore make data centers a strategic asset. But strategic infrastructure demands strategic governance. We should neither fear data centers reflexively nor accept them uncritically. We should build them rationally, transparently and responsibly, and insist on seeing the numbers.

---

Agus Hasan is a professor at the Norwegian University of Science and Technology (NTNU). The views expressed are his own.

Add as a preferred source on Google +