Google pushes water standards amid data center backlash
Facing mounting scrutiny over data center water consumption, Google on Wednesday released a set of guidelines it says should become the industry standard.
Why it matters: Communities across the U.S. are increasingly pushing back against new data centers, often citing concerns about water use alongside rising power prices, local air pollution and noise.
- Google argues that better practices — and more transparency — can help ease those fears.
Driving the news: Google’s framework calls for:
- Returning more water to local watersheds than its data centers consume by 2030.
- Avoiding water-intensive cooling in more water-stressed regions.
- Helping fund local water infrastructure upgrades.
- Pursuing alternatives such as reclaimed wastewater.
- Disclosing water use annually.
Reality check: None of these commitments are new on their own.
- The announcement largely packages together practices Google says it increasingly follows already — while turning them into a formal framework the company hopes others also adopt.
By the numbers: In 2024, Google consumed 7.2 billion gallons of freshwater and replenished approximately 4.5 billion gallons of water, which is roughly 64%.
What they’re saying: “There’s so many data center developers, and many of them are not doing it the right way, so people’s concerns are legitimate,” said Bikash Koley, vice president of global infrastructure at Google.
- “But there is also a lack of information, and water is one of those where lack of information always breeds distrust.”
Zoom out: Google joins other tech giants including Microsoft, Amazon and Meta, which have over the past several years announced goals to better manage their water consumption at their data center operations.
- While those efforts have largely focused on company-specific targets, Google is positioning its guidelines as a framework it hopes the broader industry will adopt.
How it works: Data centers need cooling because the chips running AI generate enormous heat.
- That cooling happens in two main places: first, close to the chips themselves; and second, across the broader building.
- For the hottest AI chips, companies are increasingly using liquid cooling, which moves heat away from the chips through sealed pipes. Google says its closed-loop systems use very little water because it’s continuously recirculated.
Yes, but: That heat still has to leave the building. The main tradeoff is between water and power.
- Evaporative cooling uses water to carry heat away and can require less electricity, while air cooling uses little or no water onsite but can require more electricity.
- Air cooling consumes on average 10% more energy than evaporative cooling, and roughly twice that on a hot day, said Koley.
- “It becomes a tradeoff between reducing stress on the grid versus reducing stress on the watershed,” Koley said.
State of play: Roughly two-thirds of Google’s data centers use evaporative cooling, while the remaining third is a combination of air-cooled or using recycled, non-conventional water resources, said Ben Townsend, head of infrastructure and sustainability at Google.
Between the lines: Google is making a more nuanced argument than many data center critics.
- The company argues evaporative cooling can be the better environmental choice in places where water supplies aren’t under stress because it reduces electricity demand.
Case in point: Google officials pointed to a new data center in India using air-cooling technologies, and the American Southwest generally as examples of where their due diligence into the local water supplies compelled them to use less-water intensive cooling methods.
What we’re watching: Google executives declined to predict the company’s future water use, saying local conditions heavily influence what cooling methods are deployed. Its 2025 numbers are coming out in a few weeks.
The bottom line: Google says that the share of its data centers using air cooling is rising, a sign that water concerns are increasingly shaping how companies design the infrastructure behind the AI boom.
