Water: the hidden link in AI's growth story
Power and chip supply dominate the conversation about what could slow AI's growth. Water gets far less attention, but it may matter more.


By Josh Magnuson Executive vice president and general manager, Global High Tech, Ecolab
Ask what could stall the growth of AI, and the answer usually comes back as power or chip supply. Water rarely enters the conversation, yet it sits underneath all of it.
Put bluntly, without water, we don't have AI. We can't cool the turbines that generate the power. We can't make the microchips that feed the data centres. And we can’t cool the high-performance computers that keep data centres running effectively.
Water is integral to every step of the AI value chain, yet remains one of the least-discussed parts of the story. Part of the reason is that water is still treated as an unlimited resource, rather than a vital asset for companies and communities alike. As water and energy are uniquely local resources, managing them at the scale of AI requires a different approach. Across chip manufacturing, power generation and data centres, a better model exists: one that relies on managing water with greater circularity, through reuse, recycling and next-generation liquid cooling technologies.


Water is integral to every step of the AI value chain, yet remains one of the least-discussed parts of the story.

Power runs on water
Power generation depends on water more than most people realise. Traditional thermal, gas, geothermal and nuclear power rely on water to cool the reactors or turbines spinning from steam. Hydroelectric power, by definition, uses water. Even solar and wind power, generally seen as clean alternatives, require significant water as part of the production process.
Across the entire spectrum of power, water is the constant that makes the output possible. That same output goes into making chips and running data centres, and requires reliable efficiency to meet the growing demands of AI.



With chips, the water requirement becomes more specific and demanding.

The purest water on earth
With chips, the water requirement becomes more specific and demanding. Each layer printed on to a silicon wafer requires polishing and rinsing. The AI chips made today, with the smallest features and sizes, and the most complex capabilities on the market, require hundreds of these steps to produce.
Chip manufacturers do not use ordinary water. They use ultrapure water, filtered down to parts per trillion of particulate—arguably the cleanest water on earth. As the name suggests, the quality of ultrapure water helps to maintain the integrity of delicate circuits on the chips, making ultrapure water a critical production asset that must be managed strategically.
Producing ultrapure water takes real energy and resources, yet many processes treat water as a linear resource to take and dispose of. That's part of why the current state of the industry falls short: while every chip manufacturer is unique, traditional operators recycle less than 40% of that water, discharging the rest into municipal treatment facilities. Pushing the recycling rate to 60-80% would not reduce water usage, but it would cut how much fresh water needs to be drawn from a river or reservoir, so chip manufacturers can uncouple their growth from local resource constraints and scale faster.
That matters because chip manufacturers need to grow in locations with complex local water requirements, many of which are water-stressed regions. At every new facility, a water management plan needs to be part of the design from the start, rather than added once the site is already chosen. The logic runs in both directions. Without chips, there are no data centres. Without water, there are no chips. Left unmanaged, water becomes the physical constraint that decides whether AI can scale at all.
Solving the most visible problem
Data centre cooling is the most visible water use in AI infrastructure, and is also the one problem closest to being solved. Older data centres were dependent on air cooling models to remove heat from high-performance cooling. As AI computing workloads have grown more dense, traditional air cooling can no longer keep pace. The model has shifted to manage water circularly, whether that’s through reuse and recycling or next-generation direct liquid cooling technologies. When designed strategically, cooling technology can help to reduce energy and water consumption, as well as to handle higher heat loads. Modernised data centres can run on roughly as much water as a car wash uses in a single day. At some of the most efficient facilities, the people on site use more water than the cooling systems themselves do. The technology for this already exists. It's now a matter of getting the momentum to catch on, so this becomes the standard, rather than the exception.



We believe the companies that lead and win in the AI era will be those that prioritise growth, water and energy together.

Growth and impact: connected outcomes, not trade-offs
Sustainability and growth aren't separate stories. A company has to do right by its business and by the communities where it operates. Increasingly, those two goals point in the same direction. Meeting or beating earnings targets while improving environmental performance is not a trade-off. It's simply the model for good business.
The technology exists across power, microelectronics and data centres to use less water and operate more efficiently. There is a business case for treating water risk with the same seriousness as energy risk, and as a financial and strategic question, rather than a reporting line item. The result isn't just faster permitting. It's high-quality water to meet the growing demand for chips, power and data centres without creating significant strain on the surrounding communities. Treat water as a critical input—a raw material on par with silicon or copper—rather than a waste product, and the entire conversation shifts.
Doing it right the first time
The race to build AI infrastructure is accelerating worldwide. If industry doesn't get it right from the start, the impact won't just be environmental; it will be economic too. AI is moving fast. Now it is up to us to help the infrastructure behind it evolve just as quickly. The key to AI growth is water circularity: reducing, reusing and recycling water, rather than disposing of it as an expendable resource. We believe the companies that lead and win in the AI era will be those that prioritise growth, water and energy together. Fortunately, it is possible to pursue all three.
Views or opinions expressed are those of the author and any individuals cited, and do not necessarily reflect those of Economist Enterprise or any other member of The Economist Group.