Picture this: a child opens a laptop, types a question into an AI tool and within two seconds receives a beautifully organized answer. It feels like magic. However, behind that two-second miracle sits a chain of massive server buildings humming with energy and cooled by millions of gallons of water that most of us never think about.
As a classroom teacher, I watch students use AI tools like pencils. These tools are genuinely remarkable. However, new research is pulling back the curtain on what powering that convenience actually costs and the numbers are astounding enough that every parent and educator deserves to know them. We call it “the cloud” a word that sounds airy and weightless. The reality is far more grounded. Every AI query travels to a real building in a real city, powered by real electricity and cooled by real water. By 2030, AI servers in the U.S. are projected to consume up to 1,125 million cubic meters of water annually. This is enough to fill 450,000 Olympic-sized swimming pools. Carbon emissions could hit 44 million metric tons per year, equivalent to 10 million extra cars on our roads. Major tech companies have set net-zero goals, but experts warn those targets are unlikely without dramatic change. There is also a “rebound effect”: as AI gets cheaper, we use it more — driving total environmental strain higher even as devices become more efficient.
Photo caption: Water consumption power AI servers
Where are these servers located?
Where a server is located changes everything. Hot southern states like Florida require far more energy and water just to keep machines cool. Meanwhile, states like Montana and South Dakota with cooler climates, abundant wind energy, and lower water stress are far better suited for data centers. The electricity source matters too. A coal-powered server can carry a carbon footprint ten times higher than one running on renewables. Some cities are already pushing back. For example, parts of the Netherlands placed suspensions on new data centers due to power concerns and similar debates are emerging across the American Southwest.
So how can we stop this?
As of early 2026, resistance has evolved from simple skepticism into organized protests, lawsuits, and "pro-human" marketing strategies. These efforts are gaining traction as public anxiety regarding AI’s societal, economic, and environmental impact grows. What can we do at the mean time? We can raise and teach ethically, responsible and environmentally conscious digital citizens.
3 Steps for Raising Responsible Digital Citizens
Practice Digital Intentionality: Before every query, ask: “Does AI add something here, or can a book or basic search do the job?” Skipping unnecessary requests genuinely reduces strain on power grids and water supplies.
Build Infrastructure Literacy: Help children understand the internet is physical. Watch this ABC report on how one community is being impacted by data centers for a powerful classroom example.
Turn Off Background AI Features: Disable “always-on” assistants when not needed. On Google Search, switch off AI mode if you don’t need it. It takes 30 seconds.
A rich teaching moment is asking our kids the following questions: Where was the server that answered your last AI question? What resources did it use today? The students in our classrooms today will inherit the infrastructure decisions being made right now. The best thing we can do is make sure they understand the full picture, not to burden them, but to empower them.
You can access the full research article titled "Environmental impact and net-zero pathways for sustainable artificial intelligence servers in the USA" via the following link here.