A recent marketing campaign featuring former NFL star Jason Kelce joked about using human urine to cool overheating AI data centers. While the commercial was purely satirical, it accidentally highlighted a very real industry shift. According to TechCrunch AI, data center operators are increasingly turning to alternative water sources, including treated wastewater and sewage, to offset the massive environmental footprint of AI models.
The Scale of the Cooling Crisis
The generative AI boom has created a severe thirst problem for the tech industry. Evaporative cooling, where hot air passes through water to remove heat, is standard practice for keeping servers online. But the sheer scale of modern compute is staggering.
In Loudoun County, Virginia, home to over 250 data centers and a major hub for global internet traffic: facilities collectively consume roughly 460 million gallons of water every single day. Currently, 57% of that comes directly from potable drinking water supplies. As AI models grow larger and require more processing power, relying on municipal drinking water is rapidly becoming unsustainable.
The Recycled Water Solution
To reduce the strain on local drinking water, the industry is pivoting hard toward recycled water. This involves treating municipal wastewater and sewage using membrane bioreactors, reverse osmosis, and ultraviolet light until it is safe for industrial use.
Bruno Pigott, executive director of the WateReuse Association, explains that raw wastewater contains urea, salts, and bacteria. If poured directly into a cooling tower, the hot evaporation process would create severe mineral deposits, not to mention a terrible smell. The water must be intensely purified first.
When done correctly, however, the results are highly effective. Dr. Greta Zornes, a water reuse practice leader at engineering firm CDM Smith, notes that demand for these systems has skyrocketed. Her daily engineering work is now almost entirely focused on developing recycled water pipelines for data centers.
The Infrastructure Bottleneck
Pivoting to recycled water requires significant physical infrastructure, creating several distinct challenges for the AI industry:
- Location constraints: Data centers are often built in rural areas where land and power are cheap. However, rural wastewater treatment plants simply do not process enough volume to meet a hyperscale data center’s demands.
- Buildout delays: Constructing the necessary pipelines and treatment facilities to transport and process millions of gallons of water takes years, lagging behind the rapid pace of AI deployment.
- Capital requirements: Municipalities rarely have the budget to upgrade their water treatment facilities to industrial scale without outside help.
Big Tech as Infrastructure Investors
This bottleneck is forcing AI giants to become municipal infrastructure investors. Rather than just buying water, tech companies are funding the facilities needed to process it. Meta, for instance, is investing at least $270 million in wastewater infrastructure projects near its data centers.
Michael Obradovitch, a vice president at Ecolab, points out that data centers are now acting as anchors for local water infrastructure. By committing capital to local municipalities, these companies secure their own cooling supplies while simultaneously upgrading community water systems.
Looking ahead, water management will become just as critical as securing GPUs or power contracts. On the policy front, advocates are pushing for a 30% tax credit to help industries scale recycled water infrastructure. For AI practitioners and facility developers, the mandate is clear: sustainable cooling strategies are operational necessities. The companies that succeed won’t just be those with the fastest models, but those that can secure the physical resources to keep them running.
You can find more details about specific wastewater technologies and proposed tax credits in the original TechCrunch AI report.