A.I. IS DEPLETING OUR WATER

 

I’ve heard that an A.I. data center is expected to be built at Marshfield, Missouri, where we were building a log cabin in 1982.  And I’ve been hearing some truly scary things about the physical problem with these data centers.  So I asked A.I. itself about it, and here’s its answers:

How much water did the A.I. industry use in 2025?

A.I. systems consumed an estimated 312.5 to 764.6 billion liters of water in 2025. This volume roughly equals the total amount of water consumed by the global bottled water industry in a single year.
Where does the water go?  Cooling Data Centers: Running intensive A.I. models generates massive amounts of heat. Data centers rely on evaporative cooling towers and adiabatic cooling—spraying water into the air—to keep servers from overheating.
      Individual Prompts: Depending on your location and the time of day, asking ChatGPT between 10 and 50 questions consumes roughly the same amount of water as a standard 500 milliliter (about 17 ounce) bottle.

Is that the equivalent of a city of nine million people running their taps for five months straight?

Yes, that comparison is highly accurate. 
The A.I. industry's 2025 water footprint matches the total municipal water usage of a mega-city like New York City (which has a population of roughly 8.3 to 9 million people) over a five-month span.
To break down how that math works:
      The City's Baseline: New York City consumes roughly 4.5 billion liters (1 billion gallons) of water every single day.
      Five Months of Usage: Over a five-month period (approx. 150 days), a city of that size consumes about 675 billion liters of water.
      The AI Match: This 675-billion-liter total sits squarely within the 312.5 to 764.6 billion liters that researchers calculated the A.I. industry used globally in 2025.
However, the viral phrase "running their taps straight" is a slight misnomer. If 9 million people literally left their household faucets open 24/7 for five months, the sheer volume of water wasted would be exponentially higher. The comparison refers to the entire daily water footprint of those 9 million citizens—which includes drinking, showering, flushing toilets, cooking, and running municipal services.

The Marshfield A.I. data center would greatly deplete the ground water there, drying up wells and springs, etc.  To save money, the industry is placing data centers in less expensive areas relying on wells and often experiencing droughts.  The damage is incalculable.

I have been playing with ChatGPT and enjoying it for some time, and many pictures in the Butter Rum Cartoon, and in our computer folders, are A.I.  (Not the writing though, except for A.I.’s answers in this post, and in my post, “A.I. on the Butter Rum Cartoon”).  In response to this terrible threat to our future survival, I will no longer be using A.I. as much as I can help it.  The A.I. images I’ve already collected I will keep and use (the damage is done), but will request no others from ChatGPT, will stop paying for ChatGPT Plus, and will disable A.I. answers to come up first in online searches.  It’s the least I can do.  I'm just a drop in the bucket; hopefully the whole world will learn of this, and put a stop to it.

If you think this information above is sensationalizing, please check it out for yourself.  Here are some links:

For more details on the environmental toll of AI, you can explore the full breakdown on Ekopak or check out the academic peer-reviewed study available on ScienceDirect.

 You can read the original media coverage tracking this footprint on Project Censored or look further into how tech infrastructure strains regional resources via the Brookings Institution.