OK, Can We Actually Cool Data Centers With Our Pee?

Background and Context

Jason Kelce, the Kansas City Chiefs star center, recently joked in public that people should cool data centers with urine instead of scarce potable water. On its face the remark reads like a celebrity's throwaway quip, but it latched onto one of the tech industry's most sensitive pain points. As AI compute demand surges, the water consumed by data center cooling has become a genuine constraint, and in water-stressed regions that consumption now directly competes with residents' daily drinking water.

The underlying problem is physical. Chips generate enormous heat during high-speed computation, and without timely removal servers throttle or shut down. Modern facilities rely on air or water cooling, with water systems split into primary loops that contact the equipment directly and secondary loops that reject heat through cooling towers. Secondary loops lose water to evaporation and blowdown, drawing steadily on potable supply.

For a large, hundred-megawatt campus, daily cooling water use can reach millions of liters—on the order of a small town's consumption. As AI training clusters expand exponentially, that figure climbs with them, transforming water from an ignored overhead line item into a key variable shaping where facilities can be built and how fast.

Deep Analysis

Kelce's joke, absurd as it sounds, points at a real logic: water scarcity is pushing operators to seek an abundant, uncontended replacement liquid. Technically, urine is not impossible as a cooling medium, but it faces several hard engineering barriers. First is hygiene and biofilm. Urine contains urea, salts, and organic matter that readily foster bacterial growth and biofilms clinging to heat sinks and channels, degrading heat-exchange efficiency and even clogging pipes.

Second is corrosion and scaling. The salts and chemicals in urine attack metal pipes and radiators, shortening equipment life and raising maintenance costs. Drinking water, while still requiring treatment, at least lacks these high concentrations of organics and pathogens. Third is public acceptance. Even if the engineering were solved, convincing operators and communities to accept urine-based cooling means clearing a substantial psychological barrier in marketing and public communication.

The more engineering-sound path is probably not raw urine at all, but treating it as wastewater—disinfecting and blending it with processed reuse water to reduce reliance on clean drinking water. This reframing turns a gross joke into a plausible resource-management strategy.

Industry Impact

The remark drags data center water pressure into the spotlight and reshapes the sector's competitive logic. Historically, site selection prioritized power cost, network latency, and climate, with water a secondary concern. Now that cooling consumption is impossible to ignore, water is becoming a variable on par with electricity. Regions with abundant supply or mature reuse infrastructure gain an edge in attracting new compute, while water-scarce areas face expansion bottlenecks driven by cost and water politics.

This pressure is forcing cooling innovation. The industry is exploring liquid cooling, immersion systems, free cooling, and wastewater recovery to minimize dependence on fresh water. Leading operators have begun building closed-loop systems and reuse facilities, pushing recycle rates toward near-zero discharge. In the future, water-management capability will likely become a key measure of an operator's long-term competitiveness.

That innovation also opens new markets, from high-efficiency heat-exchange materials to intelligent water-management systems, representing a meaningful incremental segment worth watching.

Outlook

Several signals warrant attention. On policy, rising water stress may prompt regulators to impose stricter metrics or fees on data center consumption, directly affecting the industry's cost structure. On technology, whether reuse-blended cooling and zero-discharge closed-loop systems can prove economic and reliable at scale will determine whether unconventional sources like urine ever enter real engineering consideration.

On the customer side, large cloud vendors and AI firms' commitments to sustainable operation will exert top-down pressure, pushing operators to invest more in water management. Kelce's joke may never become reality, but the problem it highlights is real and urgent. As each round of AI compute expansion consumes vast quantities of water, balancing technological progress with resource sustainability has ceased to be a distant ethical debate and has become a concrete choice every operator must face.

The absurd remark, in its own blunt way, reminds the entire industry to confront an issue long ignored yet increasingly impossible to avoid.