Today, pressing a button to have an algorithm write us an email or design a creative image has become an almost automatic gesture for millions of users. However, this apparent digital ease hides a colossal physical machine that operates in the shadows, far from the screens, and requires a global technological infrastructure with a voracious appetite for resources which we are only just beginning to fully understand.
The United Nations has sounded the alarm with a recent report that makes it clear that AI is not just code floating in an ethereal cloud, but a system with a leaden footing that devours energy, water, and territory. After each interaction, a supply chain is activated, ranging from the extraction of rare minerals to the massive cooling of servers, generating a an environmental footprint that can no longer be ignored if we want to stay within the planet's sustainability limits.
The energy scale of data centers in Europe
If we considered all the world's data centers as a single nation, their electricity consumption would place them eleventh globally, just behind powerhouses like France. The growth is so massive that in places like Ireland, these facilities are already They consume 21% of the total electricity of the country, exceeding even the spending of all urban households combined, which has forced the authorities to curb new connections in order to avoid collapsing the network.
In Spain, the Ministry for Ecological Transition has already begun implementing legal frameworks to hold these data centers accountable for their efficiency. European regulations are becoming increasingly strict, requiring that by 2026 new centers demonstrate a much more optimized use of electrical power contracted, preventing the digital boom from ultimately increasing the bills of ordinary citizens or compromising the national supply.
It's not just about how much energy they use, but when and where. Projections indicate that by the end of the decade, AI will account for almost 40% of the energy consumption of these centers, representing a exponential leap in demand for renewablesThis puts governments in a bind, as they must balance the ambition to be technology hubs with the commitment to reduce carbon emissions in record time.
Beyond CO2: the drama of water and soil
We tend to focus on the smoke from chimneys, but AI has an insatiable thirst. To prevent the chips from melting while processing our requests, trillions of liters of water are needed for cooling. It is estimated that in just a few years, the water footprint of this technology will be enormous. will equalize basic drinking water needs of the entire population of sub-Saharan Africa, a fact that sends shivers down your spine when you think about the recurring droughts that plague southern Europe.
Furthermore, land use is another factor that often goes unnoticed in public debates. The area required to house these infrastructures and the power plants that supply them will far exceed the size of large metropolitan areas, which represents a added pressure on land use agricultural and natural, sometimes displacing local communities who see their resources prioritized to feed foreign algorithms.
Often, these environmental burdens are not shared fairly, as while the economic benefits flow to large headquarters in the US or China, the cost of the mineral extraction and waste management The burden of electronics often falls on developing countries. This digital and ecological divide is something Europe is trying to mitigate by demanding greater traceability throughout the value chain of the hardware we use every day.
Vast differences depending on the type of task
Be aware that not all searches consume the same amount of energy: generating a short, high-resolution video can consume as much electricity as keeping an LED light bulb on for almost two days straight. This is staggering compared to a traditional Google search, which uses ten times less energy. search assisted by generative language modelsThis forces us to rethink whether we really need to use AI for trivial tasks.
Even the language we use affects our electricity meter. Experts suggest that being more concise and eliminating unnecessary polite phrases like 'please' or 'thank you' when speaking to chatbots can help. save gigawatt-hours on a global scale every year. It may sound a bit harsh, but in terms of energy efficiency, getting straight to the point is the best way to reduce our individual footprint without giving up technology.
There is also the risk of the so-called 'rebound effect,' where by making models more efficient and cheaper, we end up using them even more, negating any previous savings. To prevent the situation from getting out of hand, it is vital that technology companies Make the invisible visible and publish real data about the consumption of each of its models, allowing the user to choose the least harmful option for the environment.
Waste and the future of regulation
The speed at which AI technology becomes obsolete is staggering, which by 2030 could translate into millions of tons of electronic waste. We're talking about a weight equivalent to hundreds of Eiffel Towers in waste containing valuable but also toxic materials, requiring a much more circular waste management and responsible for not turning landfills into open-pit pollution mines.
Given this scenario, some authoritative voices in the financial and regulatory sectors are already suggesting that, if the electricity supply cannot keep pace, it may be necessary to start talking about ration AI capabilities or prioritize its use for critical sectors such as medicine or climate change. It's not about halting progress for the sake of it, but about ensuring that innovation doesn't undermine the foundations that support our quality of life and ecological balance.
The path to truly sustainable artificial intelligence lies in radical transparency, consumption optimization and a collective awareness of the physical cost of our digital requests. Only if we are able to integrating environmental responsibility into design With these same tools, we can enjoy their benefits without leaving an unpayable bill for future generations, always keeping in mind that efficiency and climate justice must be the pillars of this industrial revolution.
