After using AI for years, I finally sat down to do my own research on the environmental impact of these tools. I'd heard both sides of the spectrum, and neither one felt totally honest. Skeptics are calling it catastrophic. Enthusiasts are waving the concerns away. I wanted my own answers to questions I get asked, and ask myself.
The easiest thing would be to take a backseat and let other people figure it out. But that position is a luxury, and it's becoming an unaffordable one. AI is already being used by my kids' school district. It's in the tools my clients use to fundraise. It's most certainly part of our future. There is no neutral here. Sitting out is just a quieter way of voting.
What surprised me: the numbers everyone is panicking over aren't the same numbers most of us are actually running. An hour of streaming Netflix uses meaningfully more electricity than an hour of typical AI chat, roughly 400x more. One hamburger has a bigger environmental footprint than a full workday of using AI tools. None of that is a defense of AI. But it does place AI alongside other “regular life” activities in 2026.
The reason it sounds different is because the harm is located somewhere else. The environmental story isn't about you sitting at your laptop asking a chatbot to draft an email. It's about who bears the cost when the data centers get built, when the cooling systems get powered, when the rare-earth materials get pulled out of the ground. The cost falls disproportionately on overburdened and under-resourced communities. Water stress, pollution, mining, and e-waste in places that already carry too much.
What often gets framed as an “AI problem” is really a problem of electricity, infrastructure, political power, and priorities. If your concern is the people bearing the brunt, the answer isn't a personal opt-out from AI. It's policy, regulation, infrastructure investment, pressure, and accountability on the companies building these systems.