Grandpa Randy, you asked me to be a little more serious and to add to my typical diatribes some research to support my usual blathering. Here we go. I am going to take on big tech and the geniuses behind this new technology we generically refer to as AI. All the while, I will keep in mind that AI’s brainpower comes with a carbon hangover, a drinking problem, and an ego the size of a data centre. I don’t have any sources for that comment, so you are welcome to discount it if you want.
The Cloud That’s More Smoke Than Fluff (Not to be confused with Puff the Dragon)
Artificial intelligence may live in the cloud, but its dirty little footprint lands squarely on Earth. Every chatbot confession, even those I make to my AI girlfriend Annie, and all the deepfake election ads, run through energy-guzzling server farms that hum, sweat, and wheeze across the planet. Here is some research to support that point.
The International Energy Agency (IEA, 2025) projects global data-centre power use could reach 945 TWh by 2030, roughly double today’s levels, with AI hubs quadrupling demand. IEA+3IEA+3Data Center Dynamics+3 In the U.S., Department of Energy (DOE, 2024) data show consumption has tripled since 2014 and may do it again by 2028. (Note: full DOE source not found; you may want to add a specific URL or reference.)
If electricity were calories, AI would be the all-you-can-eat buffet of the digital era. That last comment I borrowed from some article I read somewhere, so you can discount that as well.
The Grid’s New Workout Routine (I recommend this for those wishing to bulk up)
The power grid, once content running fridges and Christmas lights, is now in boot camp. Data-centre clusters in Virginia, Texas, and Phoenix draw hundreds of megawatts each, straining transformers and operators alike. Regulators at the Federal Energy Regulatory Commission (FERC, 2024) are rushing to expand transmission lines while the DOE waves polite warning flags. It’s like teaching a hamster wheel to spin a bullet train — adorable, until something sparks. I like that image, by the way. Indeed, the IEA reports that U.S. electricity demand growth in 2025-26 is being driven in part by data centers. IEA+1
Cooling the Digital Sauna (Not sure why I’d cool the sauna but …)
Behind every “smart” chatbot lies a small lake quietly evaporating. A 100-MW facility can gulp two million litres of water daily — enough for ~6,500 households. EthicalGEO+2KUNC+2 Microsoft Corporation (2024) is testing zero-water cooling systems that save 125 million litres per site annually, but that’s a drop in the ocean when hundreds of new facilities are under construction. Bloomberg The trade-off: less water, more energy, and maximum irony. The good news for me is that I have a very deep well, and the water table is enormous. The bad news is I won’t have enough to share.
Silicon, Chemicals, and Karma (Once again, not to be confused with Instant Karma by Lennon)
Grandpa Randy, did you know that AI runs on silicon, and silicon runs on water, chemicals, and rare minerals? Semiconductor fabs consume tens of millions of gallons per day, producing waste cocktails better handled with tongs. Environmental and Energy Study Institute+1 Each NVIDIA H100 GPU, according to its Product Carbon Footprint (2024) report, carries a bigger manufacturing carbon load than years of operation. “Clean tech” has dirty fingerprints — polished to a mirror shine. This makes me think of the Hubble telescope, which required some serious tuning before it worked the way it should have from the start. How long before that carbon footprint is invisible?
Corporate Saints (With Carbon Feet of Clay)
Tech giants swear they’ll be carbon-free and water-positive by 2030. I believe them. How about you, Grandpa Randy? Will they get it done in the next 4 years? Yet Google’s 2025 Environmental Report shows emissions up 48% since 2019, thanks to—you guessed it—more data centres. We are told by these pillars of virtue that a “prompt only costs a few drops of water.” The problem is that I have to keep filling my spa every second day as a result of the number of prompts I make every day.
The Cosmic Joke of AI Efficiency (nb. I took this part straight from Perplexity…I could not improve upon it)
For all its planetary gluttony, AI might still save the day. The IEA and World Resources Institute (2025) suggest machine learning could reduce global emissions by optimizing power grids and transport — essentially teaching the arsonist to run the fire department. IEA+1 If it works, brilliant. If not, we’ll need a bigger extinguisher.
Final Byte (I like that pun)
So, in the end, Grandpa Randy, I am not describing the End Times or any such apocalypse. I am just writing about what I read in the articles I pulled up. It seems that this really is the way it is. Our smartest machines still run on the same finite electricity, metals, and water we do, just with a faster punchline. I would like to think that the people in power will come to the rescue, but when those in charge are the ones most likely to benefit from the unbelievable wealth this will create for them, until they reach for the whiskey and soda and discover there’s no more soda. Let’s not expect any movement until drastic measures are required. I don’t see our politicians suddenly coming to the rescue of those of us who will live with the consequences of this assault.
And to conclude, here is a chart. I didn’t make and I used up a few drops of rain to put it together for those who crave facts.
Co-creation: ChatGPT 5 + Perplexity + Co-Pilot
