Posted on by

AI Has a Larger Carbon Footprint than NYC, Uses More Water Than Bottled Water Industry

Reading Time: 6 minutes.

A sketch of a robot chugging water and emitting smogA study by Alex de Vries-Gao, a PhD candidate at the VU Amsterdam Institute for Environmental Studies raises some serious concerns about the carbon dioxide (CO2) emissions as well as the water usage of AI. The paper, published in peer-review journal Patterns, shows that the water consumption from AI in 2025 may have exceeded the world’s consumption of bottled water, and may have released more CO2 in the atmosphere over the year that America’s most populated city, New York. Perhaps most concerning is that another researcher, Shaolei Ren, called the findings “really conservative,” pointing out that the author excluded some variable rates of water and carbon emissions included in the supply chain and disposal of electronic equipment.

In other words, the pollution posed by AI is likely worse than even these worrying numbers.

AI’s Impact Difficult to Study

AI power and water consumption is a black box, by design. Corporations that run AI data centers could easily tell us how much power and water they’re using for AI, but to reveal that would be to invite scrutiny. It seems companies making AI products don’t want to do that. To study it, we have to look at historical trends, analyst estimates, company earning calls, and public information. The process isn’t easy, and produces estimates. If we knew how much AI was stealing from us, everyone would want to legislate it better, so companies have ample motivation to keep us in the dark.

With the way they consume power and water, that may become more literal than metaphor soon enough. Imagining rolling blackouts to ensure AI data centers have enough power isn’t too far-fetched.

Alex de Vries-Gao also runs a website, Digiconomist. There, he tracks the consumption of resources by technology. He’s previously investigated Bitcoin mining. To avoid sensationalism, his study leans more conservative, tracking power and water consumption while technologies are in use, not counting their entire lifecycle. Despite these efforts to estimate on the low side, AI has already surpassed the power consumption of global Bitcoin mining. Bitcoin is an intentionally difficult task, requiring powerful computers running constantly to produce Bitcoins. And yet the unregulated push into AI has lead the technology to use more power than Bitcoin mining.

We rarely get detailed looks into how much power AI will use, but the plans for one in the UK reveal what those data center costs might look like. That data center will be the largest in the UK, and will emit 180,000 tonnes of CO2 in a year. That’s more than 24,000 homes for one data center. The world is littered with them now, with plans to build more, and they’re driving demand for electricity and water. AI is literally stealing the resources we need to survive.

AI Using More Water Than We’re Drinking

Every year, humanity consumes 446 billion liters of bottled water. That’s a lot of water that comes from simply packaging up a natural resource. In 2023, a researcher, Shaolei Ren, was part of a study that showed AI water consumption could reach over 600 billion liters by 2027. However, AI use has exploded, and the conservative estimates from Alex de Vries-Gao’s more recent study places AI’s water consumption between 312.5 and 764.6 billion liters of water. The likely consumption over the lifetime of the hardware manufacturing and disposal or recycling are much higher.

AI is drinking more water than we package up for consumption. AI wastes more water than we drink.

AI consumes water differently than we do. When you use water, you get rid of excess through waste. Through a variety of processes, it ends up back in the local ecosystem. AI is different. It captures huge quantity of waters and does not return it where it’s taken from. This consumption depletes water from an area. In places with water stress, where water is already brought in from other areas or drought is an issue, this could deplete necessary water supplies for the living population, animals, plants, and humans.

When Meta built a data center in Newton County Georgia, a resource became more scarce. Water prices went up, wells were damaged, some “ran dry.” Some families found that they literally could not get water in their homes anymore thanks to Meta’s data center development and water usage. Reportedly, that data center may consume 500,000 gallons of water a day. That’s just one data center, consuming 182,625,000 gallons of water every year. Others that have had to put in their water requirements during the zoning process have admitted they may need as much as 9 million gallons of water per day. These are being built all over the globe. Companies prioritize electricity prices over water availability, building where power is cheep rather than places where water is plentiful. Communities pay the price. People have gone into debt and been unable to sell their homes due to the value plummeting from losing access to clean or plentiful water. Communities have had green spaces destroyed, and families have had to take loans to pay for water, all because the needs of corporations were put above the living residents of an area.

The Newton County region in Georgia is now predicted to face a water deficit, requiring more water than they can get, by 2030. A separate project planned in Phoenix Arizona will increase water stress by 32% if completed. We’re depleting our most valuable resource for AI slop.

The Millions in NYC Emit Less CO2 Than AI

NYC has over 8 million residents, with many others traveling to and from the city either as tourists or as workers who live outside of the 5 boroughs but do business here. Millions of people going about their day, every day, for 365 days, and yet we still emit less carbon dioxide than AI.

The same study that looked into the amount of water consumption that goes into cooling AI data centers also looked into the carbon emissions from the process. The results seemed to exceed that of even America’s most populated city.

NYC emits a whopping 50 million tons of CO2 every year. It’s a lot, and we have a lot to work on to fix it. The city should be greener, literally, with more tree-lined streets, wider sidewalks, and more green spaces and parks in every neighborhood. We should put bike lanes on more roads, bus lanes on more routes, replace intersections that lead to traffic jams and idling with traffic circles, and invest in more spaces for parking to reduce driving around for spaces. We could invest in our trains, invest in streetcars, and otherwise increase public transit reliability and speed to make cars obsolete for most able-bodied people. Plus, we could invest in renewables again, putting wind turbines off our coast and mandating for solar panels on new construction. There’s so much NYC could do to reduce our carbon footprint, like any city in America.

One thing all Americans can do to reduce their carbon footprint is staying the hell away from AI. Because while NYC’s 50 million tons of CO2 every year is far too much, AI’s emissions could be as high as nearly 80 million tons of CO2 for 2025 alone. This study’s estimate puts CO2 emissions from AI between 32.6 to 79.9 million tons of CO2 per year. That’s around 8% of the total greenhouse emissions for every single flight on earth over a year. One of the greatest sources of pollution has a new rival.

In 2024, AI used around 9.4GW of electricity. By the end of 2025, it’s estimated that AI will use 23GW of power. That’s more than twice the power consumption from just the year before. With companies still building more data centers, 2026 will be a year of more AI power consumption than 2025. Our electricity prices will skyrocket, we’ll accelerate climate change and the droughts, fires, floods, and storms that come with it, and we’ll deplete our water.

But, hey, how else are we going to replace human labor so capitalists can make more money from more efficient AI “workers?” Won’t someone forget about the human lives and focus on the profit potential, for once?

Ditch the Slop

“The environmental cost of this is pretty huge in absolute terms. At the moment society is paying for these costs, not the tech companies. The question is: is that fair? If they are reaping the benefits of this technology, why should they not be paying some of the costs?”

– Alex de Vries-Gao, Study Author

This isn’t fucking hard. Yes, much of AI is forced upon us by these greedy companies who add this crap to our phones, our search results, our apps, and the tools we need for work. However, we can choose not to use it. You don’t need to generate a custom emoji to send to a friend. “Oh, hey, I wasted a bottle of water so I could generate this one-time use shitty emoji of me eating a burger!” Who does that? What kind of loser goes home and has a conversation with ChatGPT instead of a loved one? What kind of moron claims to love music but really just likes having random noise in their head instead of the artistry that goes behind music? I know someone who admitted to using ChatGPT to do basic math for him instead of using his calculator on his phone or simply doing it in his head. He’s a data analyst, for fuck’s sake! A data analyst who forgot how to do math because he’s become entirely too reliant on AI. I say again, what kind of person does this? Who lobotomizes themselves so AI can generate a potentially correct response? Are these people really that lazy or so stupid they don’t realize what they’re doing to themselves?

If you’re one of those AI-obsessed losers, here’s your wake-up call. You’re using an absurd amount of power, water, and creating obscene CO2 emissions because you’re too lazy to write, do math, learn a skill, or make friends, and that’s fucking sad. You had to learn to dress yourself, surely you can learn to think for yourself. Ditch the AI.

Sources: