AI is Heating the Planet

In the middle of the night, maybe  you can not sleep because your thought are running in circle and the room feels so quiet, too heavy. So without really thinking about it you take your phone and open an AI chatbot not to get your work done, not to be productive but just to talk or asking random questions or maybe to confess something you can not say out loud to anyone else. The words come easier than you expected, you type things you have never said. The AI listens without judgment, it responds with patience and for a moment  just a moment it feels like someone actually understand you. Somewhere far away in a country you have never visited a massive building full of computers just worked hard to answer you. Fans are spinning, electricity is flowing water is being pumped to keep everything from overheating. Maybe most of us now think that AI is something that lives invisible, weightless, and clean, but actually that’s not how it works since  every AI tool that we use, like ChatGPT, Google Gemini, and those kind of it, runs on physical machines called data center. These are large buildings, sometimes the size of multiple city blocks, every single day of the year, they never stop, and they consume a big amount of energy to keep running.

These data centers need electricity, every time someone uses the AI, more energy is consumed. Right now, according to the International Energy Agency, data centers around the world account for roughly 1 to 2 percent of global electricity use. That might sound small, but it’s about the same as the entire aviation industry, AI demand is growing so fast that by 2030, that number could be much, much higher. Where does all that electricity come from? In an ideal world, it would be from renewable energy like solar panel and wind turbine, but a in reality huge portion of the world’s electricity grid still coming from fossil fuel combustions, like coal and natural gas. Aside from that, AI doesn’t just need electricity, but also need water. 

Source: Li et al. (2023). Making AI less “thirsty”: Uncovering and addressing the secret water footprint of AI models

Data centers generate enormous amounts of heat, to stop the server from overheating, they need cooling systems and many of those systems use water, fresh water, the same water that people drink. Researchers have found that a simple conversation with an AI chatbot, the kind you might have at midnight when you can’t sleep, can use roughly half a liter of water, and there are so many people doing so, the way we’re building and using AI is making climate change worse at the same time. The communities who feel this most painfully are often those who had nothing to do with creating AI in the first place. Countries in Asia, Africa, and Latin America, places that contribute very little to global tech emissions, are already experiencing the worst effects of climate change, rising sea level, longer drought.

None of this means that AI is simply bad or that people should stop using it. AI has already shows real potential to improve live in meaningful ways. It helps doctor identify diseases earlier, it also accelerate scientific research, and make information more accessible to people who previously had limited access to expertise. It is being used to model climate systems and identify more efficient energy solution. The technology itself is not the villain in this story. The problem is the way it is being built, powered, and scaled, and the speed at which expansion is happening relative to the pace of the clean energy transition.

The decisions being made right now, by our governments, by technology companies, and by the public, will shape what kind of tool AI becomes. If the infrastructure supporting AI shifts decisively toward renewable energy, and if serious investment goes into reducing the water and energy intensity of AI systems, then the technology’s growth does not have to come at the expense of the climate. Some researchers are already working on making AI models more efficient, so they require less computing power to perform the same tasks. There is also growing discussion about whether regulators should require data centers to meet environmental standards, in the same way that factories and power plants are subject to emissions rules.

AI is not going away, we probably wouldn’t want it to, since its already helping and improved millions of lives, and its potential to help solve some of the world’s biggest problems including climate change. The way we build AI and regulate AI will determine whether it becomes part of the solution or just make the crisis even worst.

Keywords: AI, Electricity, Water, Technology, Heat, Computer