If you’ve been holding off on using AI because of the environmental impact, the data tells a very different story than what’s been spreading online. Here are the real numbers, properly sourced.
Why I Wrote This
I started this account because I don’t want women to get left behind in this AI era. The people who use AI the most are getting promoted, building careers, starting side hustles, making more money.
One of the biggest reasons I hear from people who haven’t started: guilt about the environment. So I went and pulled the actual data — from primary sources, not viral tweets. Here it is.
The International Energy Agency — the most respected energy research body in the world — published Energy and AI in April 2025. Direct from the report:
1.5%
Of world electricity used by data centers in 2024 (415 TWh).
Source: IEA, Energy and AI
~3%
Projected data center share of world electricity by 2030 (~945 TWh).
IEA projection
50%
Growth in AI-focused data centers in 2025 (vs 17% for data centers overall).
IEA, 2025 update
Yes, data centers use a meaningful amount of energy. Yes, that number is growing. But the slice coming from individual people asking AI questions is genuinely tiny. The growth is enterprise: companies running AI on their own data, GPU training for new frontier models, big infrastructure deployments.
The most-shared comparison online is “one AI question = one second of microwave.” That’s roughly right — but most people are crediting it to the IEA, which never said it. Here’s where the number actually comes from.
Source 01
Epoch AI — ~0.3 Wh per AI query
Epoch AI, an independent research org, found a typical AI query uses about 0.3 watt-hours. A 1,000W microwave running for 1 second uses ~0.28 Wh. The math lines up. Hannah Ritchie at Our World in Data popularized the comparison.
Source 02
100 questions per day = ~90 seconds of household electricity
100 × 0.3 Wh = 30 Wh total. The average US home pulls about 1.2 kW. So 30 Wh / 1.2 kW = about 90 seconds of normal household use. That’s your daily AI energy footprint — a minute and a half of your home running normally.
Source 03
The IEA actually debunked the inflated numbers
Older viral claims pegged AI queries at 3 Wh per question. The IEA report calls those numbers roughly 90× too high. So if anyone is sending you a graphic with the old 3 Wh number — that math has been formally rejected by the most credible source on energy.
The IEA is clear about where the energy use is actually scaling: GPUs running enterprise AI workloads. “Accelerated servers” (the GPU-heavy systems training and running large models) account for about half of the net data center energy increase through 2030.
Translation: a hyperscaler training a frontier model for 3 months uses orders of magnitude more energy than every casual user combined for that same period. The aggregate data-center number is real, but the per-person guilt is misplaced. Your AI use is a rounding error compared to the enterprise side of the equation.
If environmental impact has been your reason to wait — the data doesn’t back the guilt. Here’s the simplest start.
01
Try Claude on the free plan
Go to claude.ai. Free plan, no credit card. Have a real back-and-forth conversation. The point is just to see what’s good about it.
02
Use it for one week on one job task
Pick the most boring or repetitive part of your work. Use Claude for that — only that — for 7 days. Don’t try to do everything. Get good at one thing.
03
Go deeper from there
Once you’ve used it for a week, you’ll know if it’s worth more time. If yes, the rest of the guides on this site walk you through every direction you can take it next.
Honest limitations
Training new frontier AI models is energy intensive — that’s where most of the cost is. But that’s a company-level concern about whether to build the model, not a user-level concern about whether to use one that’s already trained. Once a model exists, your individual queries are tiny. And the industry is racing to put data centers near renewable energy — the IEA notes a clear push toward solar, wind, and geothermal siting.
The takeaway
The guilt people feel about their personal AI use is not backed by the data. Don’t let a misattributed graphic keep you from a tool that could change your career. Use it intentionally. Use it for things that matter. And give yourself permission to start.
For Your Job
If you’re ready to set up Claude for your specific job — with custom skills, connectors, and automations built around the work you do every day — I built a bootcamp just for you.
Start the Weekend Bootcamp →Go Even Further
If you’re looking to go even further, join mine and my husband’s community group where we give you all the AI agents and systems running our businesses.
Join the Community →© 2026 Mariah Brunner. All rights reserved.