Data center water use is why AI has a location problem

Data center water use determines where you build AI facilities. These centers consume large amounts of water to keep their servers cool, prompting communities to push back against their construction. As demand for AI explodes, water availability is one reason AI has a location problem.

According to reports on Programs.com, the US has nearly eight times as many data centers as any other country, with 4,542 as of July 2026. Each of these facilities needs water to run. As their numbers increase, the strain on local water supplies is becoming hard to ignore.

The local community and tech communities are competing for water. Given this tension, many people don’t want data centers built in their locality.

How Much Water Do AI Data Centers Use?

AI data centers use a lot of water to cool their processors and prevent overheating. According to the Environmental and Energy Study Institute:

  • Each 100-word AI prompt uses roughly 519 milliliters
  • A medium-sized data center can use up to 110 million gallons of water a year
  • A large data center can use up to 1.8 billion gallons a year
  • US data centers together were estimated to consume about 163.7 billion gallons a year

These figures are high. Because of them, data center water use is becoming such a serious environmental and community concern.

How Are Data Centers Cooled?

The type of cooling method you use in your data center affects how much water they use. Here are ways in which data centers are cooled:

Evaporative Air Cooling

Evaporative air cooling is a traditional approach. It uses cooling towers that rely on evaporation to dissipate heat. While it uses relatively little electricity, you’ll use a lot of water, since large volumes evaporate.

Liquid Cooling

Liquid cooling is a newer data center cooling method. It delivers liquid coolant directly to the chips via direct-to-chip or immersion cooling, in which components are bathed in a heat-absorbing liquid.

With this method, a lot of heat is efficiently captured, and you use less water than with evaporation. However, it’s complex and costly to invest in this method.

Why Does AI Have a Location Problem?

AI data centers have a location problem. Here are some reasons why:

Water-Stressed Regions

Many data centers are easy to build in warm, dry regions like the American Southwest because of the availability of land and solar energy. However, these areas also struggle with water.

If you own a data center in a hot weather region, you’re likely to use more water to cool it. Building your AI infrastructure in a drought-prone area is a source of growing conflict.

Community Pushback

As residents learn how much water these facilities use, many are pushing back. Communities worry about data centers competing for the limited water supply with their:

  • Homes
  • Farms
  • Businesses

This struggle is influencing where new AI infrastructure gets approved.

The Climate Trade-Off

Every location involves a balance of water, power, and climate. If you’re in a cool, wet climate, you’ll reduce your data center’s cooling needs, but you may not get cheap power or fast grid access. In a sunny climate, you can save money on solar energy, but you may struggle with water access.

Where Does a Data Center’s Water Footprint Come From?

A data center’s water footprint is bigger than the water it uses on-site. The footprint comes from:

  • On-site cooling: Water used directly to keep servers from overheating
  • Power generation: Many power plants that supply electricity consume a lot of water to produce steam
  • Chip manufacturing: Uses ultrapure water to clean and produce the semiconductors inside servers

It’s often easy to overlook the hidden parts of this footprint. As a result, you can understate your data center’s impact on water infrastructure.

How Can Data Centers Reduce Their Water Use?

If you own a data center, you can reduce your water use for the benefit of the community you’re in. Here are some ways you can do it:

Use Recycled and Non-Potable Water

Data centers can use recycled, reclaimed, or non-drinking water for cooling instead of fresh drinking water. This conserves precious potable supplies. Many operators are choosing to use recycled and non-potable water to reduce their strain on local communities.

Adopt Efficient Cooling

You can start using liquid cooling and then track its efficiency. When tracking, use metrics designed to measure water per unit of computing. It’ll help your data center steadily cut consumption.

What you measure gets managed. Additionally, you can pair efficient cooling with sustainable technology to shrink the industry’s footprint.

Invest in Better Water Infrastructure

Delivering, recycling, and managing water at scale require robust piping and infrastructure. Durable, leak-resistant systems built with methods such as HDPE fusion help move and reuse water efficiently, with minimal losses.

Frequently Asked Questions

Why Don’t Data Centers Use Air Cooling to Avoid Water Usage Completely?

Some data centers do, and pure air cooling can use almost no water. However, it isn’t practical everywhere. Air cooling without evaporation struggles to keep up in hot climates.

If you have high-density AI chips, which generate intense heat, it can be hard to rely on air cooling alone. You’ll also use more electricity, which will lead to an energy problem.

What Can Communities Do About a Data Center’s Water Demands?

Communities have more influence than they may think. Local governments can:

  • Require water-use disclosures
  • Negotiate consumption limits or recycled-water commitments
  • Factor water impact into permitting decisions

As a resident, you can participate in public hearings and push for transparency about how much a proposed facility will consume.

How Does AI Water Use Compare To Everyday Activities?

AI water use adds up fast. While a single AI prompt uses only a small amount of water, the world enters billions of prompts. As a result, this situation increases water usage.

Sustainable Data Center Water Use With Efficient Cooling

Data center water use affects local communities. As a result, they may push back against having these data centers in their local areas. To avoid being pushed out, you can invest in recycling water and efficient liquid cooling to reduce water use.

For more guidance, subscribe to our newsletter.