Artificial intelligence has become one of the most talked-about technologies of the 21st century. Tools like ChatGPT feel light, quick and helpful, but behind every answer lies a huge machine infrastructure that consumes vast amounts of energy and natural resources. While AI can support society in many ways, its environmental footprint is becoming impossible to ignore. According to recent research and environmental reports, AI is not a neutral technology and its ecological costs are growing fast.

AI Is a Huge Energy Consumer

At the heart of every AI system are data centers, which are massive facilities filled with powerful servers that run continuously. These facilities aren’t small or efficient like your laptop, they require immense computing power to process and answer questions, especially for large language models such as ChatGPT. Each prompt you send triggers many complex operations on specialized hardware, which draws electricity constantly. A Greenpeace supported study found that the electricity demand of AI data centers is projected to increase eleven-fold by 2030 compared to 2023 levels. This dramatic rise is not because data centers used to be small, but because the scale of AI use keeps exploding. What was once cutting-edge is now everyday: billions of users and countless prompts add up. And this surge could seriously threaten global climate goals. To put this into perspective, even without individual figures for each prompt, the world’s AI infrastructure as a whole is expected to consume hundreds of billions of kilowatt-hours annually that is comparable to the total electricity use of entire countries. And that’s just electricity for computing.

Water: The Invisible Environmental Victim

Most people assume AI only affects electricity grids. But the truth is far more complex and alarming. The same cooling systems that make energy use possible require enormous amounts of water. Servers generate heat, and the best way to cool them efficiently at scale is often with water-based systems. These systems can evaporate water in huge quantities, especially in warm climates, intensifying stress on local water supplies. Greenpeace research reveals that the water needed just for cooling AI servers could grow from tens of billions of liters today to over 660 billion liters by 2030 which is nearly four times higher than current demand. Even more concerning: AI-dedicated centers use roughly twice as much water as regular data centers. This means AI growth doesn’t just strain energy grids , it strains freshwater systems too, especially in water-scarce regions.

Carbon Emissions Still Climb

You might think that as renewable energy becomes more common, AI’s carbon footprint would naturally shrink. But even when assuming a growing share of clean power, AI’s total emissions are still expected to rise sharply. The Greenpeace report models emissions increasing from around 29 million tons of CO₂-equivalent in 2023 to 166 million tons by 2030. Most of that increase comes from the massive expansion of computing infrastructure rather than a dirtier power mix. This means AI could be a significant driver and not a reducer of global emissions if no corrective action is taken. That’s especially worrying because AI is often marketed as a tool that will help the climate, when in reality its underlying infrastructure could be pulling in the opposite direction.

A Paradox: AI Helps Solve Environmental Problems — but Also Creates Them

Environmental advocates in Switzerland highlight AI’s “paradox”: it can help solve environmental issues, yet it creates significant environmental problems. For example, AI can improve energy efficiencies, optimize transportation routes, or accelerate climate research. These are real benefits and can help sustainability goals when used wisely. However, the scale of these benefits doesn’t automatically outweigh the costs. Just because AI could help save energy doesn’t mean it does in every case. Most AI activity today is driven by convenience, entertainment, and economic growth and not climate protection. And as AI use grows, so does overall energy consumption.

Why This Matters

Many people are surprised when they learn that everyday AI usage ties back to vast amounts of energy and water. It’s more than a digital technology, it’s a physical system with real environmental consequences. If AI’s growth continues without strong regulation, transparency, and shifts toward truly renewable power and cooling methods, its environmental costs could outweigh its benefits.

Conclusion: A Tool With a Footprint

AI isn’t fundamentally bad. But it is resource-intensive, and the environmental impacts are real. Every prompt doesn’t just compute an answer it relies on power from somewhere, water from somewhere, and technology produced at a cost. Balancing AI’s promise with its physical footprint requires honest discussion, better reporting from tech companies, and serious commitments to sustainability. Only then can we decide whether AI will be a tool for a greener future or another drain on our planet’s limited resources. In my opinion, AI tools are very important nowadays, but it is crucial that users understand how these tools work and how much energy they consume. I think it is important to use AI tools and embrace technology, but do so thoughtfully and use AI only when it is really needed.

Zürcher Tanja

Sources:

Greenpeace Germany
Artificial Intelligence: Energy Consumption and Environmental Impact
https://www.greenpeace.de/ueber-uns/loesungen-finden/kuenstliche-intelligenz-energieverbrauch-und-umweltauswirkungen

Sustainable Switzerland
Artificial Intelligence – Curse or Blessing for the Environment?https://sustainableswitzerland.ch/artikel/kuenstliche-intelligenz-fluch-oder-segen-fuer-die-umwelt-id.3503

OpenAI. (2025). ChatGPT (version 5.1).
htps://chat.openai.com
• Language support
• Brainstorming

Image source:

Sustainable Switzerland. Google’s data center in Middenmeer, Netherlands, is powered by renewable energy.