Ok, can we actually cool data centers with our pee?

Former NFL star Jason Kelce made a tongue-in-cheek suggestion that urine could replace clean drinking water for cooling data centers — and surprisingly, engineers and scientists say the idea isn't entirely without merit. As AI infrastructure strains freshwater supplies globally, unconventional cooling solutions are gaining serious attention.

Retired Philadelphia Eagles center Jason Kelce sparked an unlikely tech conversation with an offhand joke proposing that human urine could serve as a substitute for potable water in data center cooling systems. While the comment was clearly humorous, it touches on a genuinely pressing problem: modern data centers, particularly those powering AI workloads, consume enormous volumes of fresh water to manage heat.

Engineers have long explored alternative cooling liquids, and the core requirement is simply a fluid that can absorb and transfer heat efficiently. Urine, being mostly water, technically has some of those properties, though the practicalities — corrosion, sanitation, smell, and logistics — make it deeply impractical at scale.

Still, the broader conversation Kelce accidentally ignited is a valid one. The tech industry is under growing pressure to reduce its freshwater footprint, pushing companies toward recycled water, seawater, and air-cooling alternatives. Sometimes it takes an absurd joke to highlight a very real problem.

Jason Kelce, the recently retired NFL lineman turned media personality, probably wasn't expecting to contribute to the discourse on sustainable infrastructure — but his quip about cooling data centers with urine instead of drinking water has done exactly that. The joke, made in his characteristically blunt style, managed to land in a genuine debate about one of tech's dirtiest secrets: its massive appetite for fresh water.

Data centers use water primarily in cooling towers, where it evaporates to dissipate the enormous heat generated by servers. A single large facility can consume millions of gallons of water per day. As generative AI models grow larger and more computationally demanding, this water consumption is accelerating, raising alarms in drought-prone regions where these facilities are often built.

So could urine actually work? Scientifically, urine is roughly 95% water, which means it does carry heat-transfer properties. But the remaining 5% — ammonia, urea, salts, and various organic compounds — would corrode metal components, foul filtration systems, and create serious biohazard and odor challenges. Collecting enough of it at industrial scale would also be a logistical nightmare that likely negates any environmental benefit.

Why it matters: The joke, absurd as it is, exposes a real tension at the heart of the AI boom. Tech companies are racing to build ever-larger data centers to meet demand, but water scarcity is already a geopolitical flashpoint in many parts of the world. Communities in places like the American Southwest and parts of Europe have begun pushing back against data center construction precisely because of freshwater competition. The industry needs genuinely creative solutions — recycled greywater, seawater cooling in coastal facilities, advanced air cooling, and immersion cooling with synthetic fluids are all being explored seriously.

Kelce's joke, intentionally or not, is a useful mirror: if even a throwaway comment from a retired football player can generate real engineering discussion, it underscores how urgently the public and policymakers are beginning to scrutinize Big Tech's environmental footprint. The water problem isn't going away, and solving it will require ideas far less ridiculous — and far more politically difficult — than the one that started this conversation.