AI Thirst Swallows Down Drinking Water

A former NFL star tells you to mail jars of your urine to data centers. It's a joke. But is the premise as absurd as it sounds? The answer reveals how the tech industry plans to sustain its growth in a world with less water.

The campaign from Liquid Death and Jason Kelce starts with a blunt statement. "AI data centers waste millions of gallons of water," Kelce says according to TechCrunch. His "solution" is for people to drink their products and donate the resulting pee for cooling. It's satire designed to go viral. Yet it accidentally points to a real, pressing dilemma: AI's infrastructure is colliding with climate-driven resource limits.

A single cluster of data centers can swallow a small town's worth of water. Loudoun County, Virginia, a major hub, offers a snapshot. Its data centers used about 200 million gallons of recycled water per day as of 2025. Sounds good. But that recycled water only covered 43% of their daily needs. The other 260 million gallons, a majority 57%, still came from potable drinking supplies. This isn't a niche problem. A Department of Energy-funded study cited how data centers across the United States consumed 66 billion liters of water. The pressure on local water systems is already sparking opposition, with about 70% of Americans opposing new data centers in their communities.

This conflict is why Kelce's punchline lands. It frames the tech industry's thirst as a gross, public emergency. The question it leaves us with isn't about pee. It's about what happens when the clean water runs out.

Can We Just Use Dirty Water Instead?

The core of Jason Kelce's joke contains a sliver of truth. You wouldn't use urine, but data centers are already using what comes right after it.

"The Liquid Death commercial is funny and tongue-in-cheek," said Michael Obradovitch of Ecolab. "But in reality, there is a fair amount of alternative water sources already being used to a similar extent to cool these data centers."

The serious alternative is recycled water. This is wastewater and sewage that has been treated to a standard safe for industrial use, not for drinking. It passes through facilities using membrane bioreactors, reverse osmosis, and UV light.

"You wouldn’t just use pee, but you can clean it and make it into useful water, and that’s what we advocate," said Bruno Pigott of the WateReuse Association.

The benefits are straightforward.

• It conserves drinking water, taking pressure off municipal reservoirs.
• It provides a reliable, local supply that isn't subject to the same restrictions as potable water during droughts.
• For operators, it can be a cheaper, more sustainable source.

Dr. Greta Zornes of CDM Smith confirmed the trend. "We use recycled water for cooling for all kinds of industries, and we have for decades," she said. "So this is only one application, but definitely, there’s been a boom in recycled water for data center cooling."

The limitation is infrastructure. You need to be near a large treatment plant. "When data centers go out into rural areas, a lot of times the wastewater treatment plants just aren’t big enough," Zornes noted. Building that pipeline takes time and capital data center operators aren't always prepared to provide.

What If We Didn't Need Water At All?

Recycled water is a better source, but it's still water. The industry's more radical direction is to design systems that bypass evaporative cooling entirely.

One leading alternative is immersion cooling. Instead of blowing air over hot components, servers are submerged in tanks of non-conductive, biodegradable fluid. The fluid directly contacts heat sources like CPUs and GPUs, absorbs the heat efficiently, and is pumped away to be cooled in a closed loop. This system uses zero water in the actual cooling process. It's also more energy-efficient and allows for denser server packing.

Another method is air-side economization, which uses cool outside air directly. This works brilliantly in places like the Pacific Northwest but fails in hot, humid climates where you'd need energy-intensive air conditioning as a backup.

These technologies represent the real endgame: reducing the water footprint to near zero, not just substituting one water source for another. They address the root criticism that AI's environmental cost is too high for its perceived value, a tension felt acutely as models grow larger. It's a shift from finding more liquid to building smarter hardware.

Why Is Human Urine A Terrible Idea?

Let's return to the joke. Why is piping raw pee to a server room a catastrophic plan? The experts are clear.

"Pee contains all sorts of stuff. It contains salts, it contains urea, bacteria, organic matter of all sorts that can leave mineral deposits," explained Bruno Pigott. "If you just put that into a cooling tower or something else, it would require constant cleaning."

Cooling towers, a common component in many data centers, work by passing hot air through a water spray. The water evaporates, removing heat.

"Can you imagine if you just poured urine through hot air?" Pigott asked.

Even setting aside the smell and biohazard, the logistics are absurd. The volume required is staggering. Collecting, storing, and transporting millions of gallons of urine would create a waste problem far worse than the one it purported to solve. The process of treating that urine to a usable standard is precisely what wastewater plants already do at a municipal scale. Kelce's gag highlights the resource problem correctly but inverts the solution. We don't need a new bodily fluid source. We need to scale the infrastructure to treat the wastewater we already produce.

Who Pays For The Pipes?

The future of data center cooling isn't a mystery. The path is clear: use more recycled water where possible, and deploy advanced dry-cooling tech where it's not. The bottleneck isn't science. It's money and coordination.

Dr. Zornes pinpointed the delay: "There’s a lot of infrastructure that has to be built out and usually isn’t existing today... That’s one of the problems | it’s just the time that it takes to get that done."

Who funds that infrastructure? There are two emerging models.

1. Tech Companies as Anchor Tenants. Companies like Meta are stepping in. Obradovitch noted, "There’s a number of cases and examples where data centers, as part of their engagement with communities, have committed funding and capital to some of these municipalities." Meta, for instance, has committed at least $270 million to wastewater projects near its data centers. The data center becomes the economic justification for upgrading a town's entire water system.
2. Policy Incentives. Bruno Pigott is advocating for a 30% tax credit to help industries scale recycled water infrastructure. "We think it would greatly accelerate the pace," he said. This would lower the financial barrier for operators and municipalities alike.

The choice communities face is becoming starker. They can reject data centers outright, as our recent coverage on Amazon Bulldozes Rare Books for AI-Fueled Data War explored. Or, they can negotiate hard for the infrastructure investments that leave their water supply better off. The outcome hinges on whether the tech industry's need for growth can be aligned with the public's need for resource security. It's a negotiation that will play out in county zoning meetings and state legislatures, far from the viral marketing campaigns.

Impact Analysis

• AI's explosive growth is creating immense demand for water to cool power-hungry data centers.
• Public opposition is rising as data centers consume community drinking supplies, with about 70% of Americans against new ones nearby.
• The industry's sustainability and growth depend on finding new, non-potable water sources before clean water runs out.