How Much Energy Does AI Actually Use — And Is It Destroying the Planet?

Anamika Dey, editor

By TechSun News Desk | techsunnews.com | October 13, 2026 | Tech / AI / Green Tech | 6 min read

Every time you open ChatGPT and type a question, something happens that you cannot see. Somewhere in a data center — probably in Texas, Virginia or Iowa — a rack of servers lights up and draws power. Not a lot. But multiply that by 900 million weekly users and the numbers start getting very big, very fast.

So, how much energy does AI use, and is it destroying the planet? The honest answer is more complicated than either the alarmists or the tech optimists want you to believe. Here is what the data actually shows.

The Numbers — Let’s Start With What’s Real

First, some perspective. Older estimates put a single ChatGPT query at roughly ten times the energy of a Google search, about 2.9 watt-hours versus 0.3. Newer measurements suggest a typical text query is now closer to 0.3 watt-hours, though long reasoning prompts and images can use far more. For scale, 0.3 watt-hours is about what a 10-watt LED bulb uses in under two minutes.

Here is the full picture:

The big number is that last one. The IEA estimates the world’s data centers used about 415 TWh of electricity in 2024, and it expects that to more than double, to around 945 TWh, by 2030. AI is the biggest driver of that growth, even though AI servers were still a small share of data center demand in 2024. For a closer look at what a single prompt costs in electricity and water, see our guide to ChatGPT energy use.

Why Data Centers Are the Real Story

When people talk about AI energy use, they usually focus on the wrong thing. The energy used to answer your ChatGPT query is tiny. The energy used to train the AI model that answers it is enormous.

Training GPT-4 reportedly consumed around 50 million kWh of electricity — a one-time cost that powers the model for all its future use. But AI companies are training new models constantly. And then there is inference — the energy used every time someone actually uses the model — which adds up continuously across billions of daily queries.

This is exactly what is driving the debate we covered about Sanders and AOC’s push to freeze new AI data center construction in the US. Their argument: AI companies are building massive power-hungry facilities faster than the grid can handle it — and ordinary Americans are paying higher electricity bills as a result.

And with Microsoft putting AI agents directly inside Windows — meaning AI is now running continuously on hundreds of millions of personal computers, not just in data centers — the energy footprint is about to get significantly larger.

The Overlooked Water Cost of AI

Energy is actually only half the environmental story. The other half is water.

AI data centers use enormous amounts of water for cooling. Microsoft reported that its data centers consumed about 6.4 billion litres of water in 2022, a 34% rise on the year before, and that was before the current AI boom. Google’s water consumption rose 20% over the same period. Researchers have estimated that training GPT-3 alone evaporated about 700,000 litres of fresh water just to cool the servers on site.

In regions already facing water stress — the American Southwest, parts of Europe, large areas of Asia — this is not an abstract concern. It is a genuine resource conflict between AI infrastructure and local communities.

So Is AI Actually Bad for the Planet?

Here is where it gets genuinely complicated — and where most coverage gets it wrong.

AI is simultaneously one of the biggest new sources of energy demand AND one of the most powerful tools we have for reducing energy use elsewhere.

Examples of AI cutting emissions:

• In a 2016 trial, Google’s DeepMind AI cut the energy used to cool its data centers by up to 40%
• AI-powered grid management is helping utilities spot and reduce wasted electricity
• AI weather forecasting is making wind and solar power more predictable and therefore more usable
• AI drug discovery is cutting years off medical research that would otherwise require enormous lab energy

The honest picture: AI’s net environmental impact depends almost entirely on what it runs on. An AI data center powered by renewables is a very different thing to one powered by coal. Right now, most data centers run on a mix — and the race is on to shift that mix toward clean energy. Whether it gets there fast enough is the real question. The International Energy Agency’s Energy and AI report projects that data center electricity demand will more than double by 2030, which puts real pressure on power grids and on how fast clean energy can be built.

This connects to the bigger picture of how AI is changing everything about how we live and work — the environmental cost is part of that story, not a separate one.

What You Can Actually Do About It

Genuinely — not much individually. Your personal AI usage is a rounding error in the global picture. But a few things are worth knowing:

• Text queries use far less energy than image or video generation — if you are using AI for writing and research, your footprint is minimal
• Using AI tools that run on renewable-powered infrastructure helps. Google and Microsoft both publish annual sustainability reports showing their energy and water use
• Supporting companies and politicians who push for renewable data centers matters more than your individual usage
• Understanding what AI tools actually collect and how they work helps you use them more intentionally — our guide to what ChatGPT actually does is a good starting point

The deeper issue is that AI’s dark side goes beyond just energy — from data privacy to job displacement to environmental cost. Understanding all of it is how you make informed decisions about which AI tools to use and how much to trust the companies building them.

And if you are wondering whether the energy cost of AI is worth it given AI is changing how information is found online — that is exactly the right question to be asking.

FAQ — AI Energy & Environment

1. Is using ChatGPT worse for the environment than Google?

Older estimates put ChatGPT at roughly ten times the energy of a Google search per query, though newer measurements suggest a typical text query is now much closer to a search. Either way, it helps to keep perspective: even a full day of heavy ChatGPT use uses less energy than a 10-minute hot shower. The bigger environmental story is the data centers and training runs, not individual user queries.n a 10-minute hot shower. The bigger environmental story is the data centers and training runs, not individual user queries.

2. Are AI companies doing anything about this?

Some are trying. Microsoft has committed to being carbon negative by 2030, and Google has set a target of net-zero emissions by 2030. Both are investing in renewable energy and more efficient data centers. But both have also reported higher emissions in recent years as they build more AI capacity, so the gap is still a live concern. Smaller AI companies often disclose less about their energy and water use, which makes their impact harder to measure.

3. Should I feel guilty about using AI?

No — and anyone who tells you your personal AI usage is a meaningful contributor to climate change is misdirecting you. The decisions that matter are made by governments, energy regulators, and the AI companies building the infrastructure — not by individuals asking ChatGPT to help write an email. Focus your energy (no pun intended) on understanding the bigger picture and supporting the right policies, not on personal guilt about your query count.

techsunnews.com | Tech / AI / Green Tech | © 2026

You Might Also Like

OpenAI’s GPT-5.6 Is Here — But the US Government Won’t Let You Use It Yet Kimi K3 Explained: Why China’s New Open AI Model Is Getting So Much Attention Why GPUs Are Suddenly So Expensive in 2026 — And Why AI Is the Real Reason