AI Is a Water Hog: Why the Future of Data Centers Flows Through Sewage
By George S. Pullen, Chief Economist, MilkyWayEconomy
Artificial Intelligence doesn’t just consume data and electricity, it devours water. Every prompt, every model update, every whisper from an algorithmic oracle hides behind thousands of gallons of evaporated coolant. For all its sophistication, AI remains bound to a primitive, biological limitation: the need to stay cool.
In 2024, estimates from industry analysts suggested that a single hyperscale data center can draw 1.25–5 million gallons of water per day, mostly to evaporate into the air through cooling towers. Multiply that by thousands of facilities globally and AI begins to resemble not a digital revolution, but a hydrological extraction industry.
The irony is thick. We’re training AI to solve climate, food, and energy crises while accelerating localized water insecurity in the process. As compute density rises, so too does the competition between data and drinking water.
The Economic Efficiency Hidden in Waste
But there’s a quiet revolution already underway, born not in Silicon Valley but in the unsung world of municipal sewage treatment plants. These sites, long seen as the backwaters of urban infrastructure, may soon become prime real estate for the AI economy.
The logic is elegant:
Wastewater treatment facilities already handle massive volumes of grey and black water, which are non-potable but thermally valuable. That water is typically discharged after secondary treatment. Yet, before it’s released, it could pass through data center heat exchangers, absorbing waste heat while keeping processors cool.
In this cycle, AI heats water, wastewater cools AI and biogas from sewage powers the system, a literal closed-loop economy of waste and computation.
Imagine the infrastructure stack of the future:
- Wastewater in → AI computation → Heat exchange → Enhanced digestion → Biogas out → Power loop complete.
This isn’t just ecological, it’s economic. Wastewater treatment is one of the most energy-intensive municipal functions. Co-locating it with data centers creates shared utilities: electricity, heat management, and even land use. Cities can convert sunk costs into revenue streams by leasing treated water for industrial cooling.
Urban Economics in the Age of Heat
Cities of the 21st century will be defined by how well they handle heat, not just atmospheric heat but digital heat. Heat is a currency. The value lies in where it’s captured and how it’s reused.
Countries like Denmark and Finland are already pioneering “heat recycling,” using waste heat from data centers to warm homes through district heating systems. The U.S. can and must go further. The next generation of municipal utilities should merge data infrastructure and wastewater management into one integrated hydrological-computational system.
The incentive structure is clear:
- Municipalities gain new tax revenue and energy savings.
- Data companies reduce cooling costs and ESG liabilities.
- Citizens benefit from stabilized water prices and local job creation.
In essence, we’d be re-plumbing capitalism, turning waste into an asset and heat into a form of local wealth.
The Sewage Economy: A Catalyst for the 6th Industrial Revolution
The convergence of waste management, AI and renewable energy marks a quiet entry point into what I call the 6th Industrial Revolution—a period defined not by new extraction but by total system integration.
This revolution won’t be televised or streamed, it’ll be piped. Beneath our cities, a new circulatory system will emerge, where bytes and bacteria share infrastructure. Where gray water cools intelligence, and artificial intelligence powers the reclamation of waste.
It’s not glamorous but it’s profoundly scalable. The future data center will be less like a server farm and more like a biome, embedded within the living infrastructure of the city it serves.
AI may be a water hog today. But tomorrow, when computation meets circulation, it will become the heartbeat of a regenerative urban ecosystem.
George
Disclaimer: The opinions expressed herein are those of the author and do not necessarily represent the views of any affiliated entity.