Data Centers: The New Boogey Man
- Allie McCormack
- Jul 1
- 4 min read
AI MYTHS: DEBUNKED "The Data Center Water Panic, Explained"
You've seen the headlines. Data centers are "guzzling" water. Towns are going dry. Somewhere, a server farm is allegedly drinking your tap water while you wait for your turn at the faucet.
Before we get to the AI-specific freakout (and there is one, don't worry), let's deal with the bigger myth first: that data centers — any data centers, AI or not — are some kind of unprecedented water apocalypse.
Myth: Data centers are draining entire regions of drinking water.
Reality: Data centers don't consume water like a factory eating raw materials — they use it for cooling, and most of what they use either evaporates (like sweat) or gets returned to the system. Unlike many other industries, data centers don't use water as a raw material; they use it for cooling. It's the same basic principle as a swamp cooler, just bigger and cooling a server rack instead of your living room.
The Statistics: All data centers combined use about 0.05% of U.S. freshwater — that's less than the water footprint of a single mid-sized New Jersey city. And the "they're stealing your water" framing usually skips a key fact: households account for only about 1% of Americans' total water footprint, with the other 99% going to food production, electricity generation, and manufacturing.
Even in places singled out as ground zero for the panic, the math doesn't hold up. In Maricopa County, Arizona — one of the most water-stressed areas in the country — data centers use 0.12% of the county's water. Golf courses use 3.8%. Nobody's writing viral threads about golf or attacking golfers.
And when local water systems do need upgrades because of new data center demand? Investigative reporting hasn't found a single instance of data center operations raising household water bills in America — in some cases, like a notable Oregon facility, the data center's infrastructure investment kept rate increases lower than they otherwise would have been. Companies are also funding water system upgrades directly — Microsoft, for instance, invested over $40 million in Goodyear, Arizona's wastewater infrastructure as part of building there.
So: data centers use water, communities should absolutely keep an eye on that, and some siting decisions deserve real scrutiny (drought-prone areas, transparency about projected use). But "your shower is in danger" was never an accurate read on the situation.
Now for the AI-specific version of this myth, which gets extra spicy because it comes with a built-in villain: chatbots.
Myth: Every time you ask an AI a question, you're personally damaging the planet.
Reality: This one tends to come from comparisons that sound shocking precisely because they're scaled wrong — like the now-debunked viral claim that a single overseas Google data center would use a thousand times more water than the city around it. There's just one problem with claims like that: the math is completely wrong.
The Statistics: Take the most water-intensive AI training run anyone's talking about. Training the largest AI model to date consumed less water than what a single square mile of farmland uses in one year — and that's one of the largest AI data centers in the world.
Worth noting too: a lot of the scariest-sounding AI data center stories aren't actually about the data center using too much water — they're about something else entirely getting mislabeled. One widely shared story about wells drying up near a Meta facility buried the actual cause in the article body: the data center hadn't even been turned on yet — the real culprit was construction sediment, the kind of issue that could happen at any construction site, AI-related or not.
None of this means AI's water and energy footprint deserves zero scrutiny — newer, bigger facilities are real, and demand is genuinely climbing as adoption grows. A February 2026 report estimated data center potable water consumption in England alone at close to 1.9 million cubic meters a year and trending upward. That's a legitimate thing to track. It's just a very different conversation than "AI is draining the planet of water," and it's one we can have honestly without the apocalyptic math.
Bottom line: Data centers use water. So does literally every other piece of infrastructure that keeps modern life running. The question isn't "should this use zero water" — it's "is it being managed responsibly," and on that front, the industry is doing a lot more than the panic headlines let on.
Bonus: While We're At It — Energy Gets the Same Treatment
Water isn't the only thing people love to blame exclusively on AI. Energy use gets the same selective outrage.
Myth: Chatting with AI is uniquely wasteful.
Reality: Two prompts to a chatbot use less energy than almost anything else you do online. According to a Forbes breakdown of data from TRG Datacenters, an hour of Netflix or YouTube burns about 500 times more electricity than two AI prompts. Zoom calls aren't much better — an hour on a video call uses roughly as much energy as a short AI-generated video clip.
The Statistics: Here's the lineup, cheapest to priciest:
Two AI chatbot prompts: ~0.0002 kWh
Voice assistant query (Alexa, Siri): ~0.0005 kWh
One AI-generated image: ~0.003 kWh
Short email, no attachment: ~0.013 kWh
Zoom call, 1 hour: ~0.05 kWh
Text-to-video generation, 6–10 seconds: ~0.05 kWh
Netflix or YouTube, 1 hour: ~0.12 kWh
So the next time someone tells you asking a chatbot a question is "killing the planet," you can point out they're the ones bingeing four hours of Netflix that same night — and that's before we even get to the data center water usage, which is a whole separate (and, yes, real) conversation. I have a blog post on
Comments