AI does not fall from the sky—it requires vast amounts of energy and water.
Beneath Switzerland's reputation for precision and environmental stewardship, a quieter reckoning is unfolding: the invisible machinery of artificial intelligence is consuming electricity and water at a pace that strains the country's renewable ambitions and the patience of its citizens. With more than fifteen data centres under construction and projections suggesting AI infrastructure could claim 15 percent of national electricity by 2030, Switzerland confronts a question that transcends technology policy — how much of the natural world is a digital future permitted to consume? The public, it turns out, has already begun forming an answer.
- Switzerland's data centres already draw up to 8% of national electricity, and that share could double by 2030 — outrunning the country's own renewable energy expansion targets.
- Four out of five Swiss citizens are worried, and in a hypothetical energy shortage they ranked data centres last among all sectors — a striking rebuke to the AI industry's sense of its own importance.
- Water is emerging as a second fault line: seven in ten respondents fear that the cooling demands of AI infrastructure will quietly damage the ecosystems Switzerland has long prided itself on protecting.
- Transparency is the missing link — some companies publish consumption figures, but independent verification is rare, and the public is demanding legally binding disclosure of energy sources, usage, and environmental impact.
- The Federal Council has promised AI regulations by end-2026, but pressure is mounting to ensure those rules carry real environmental teeth rather than simply ratifying the growth already underway.
Switzerland is facing a crisis that has crept up quietly. Beneath the promise of artificial intelligence lies a hunger for electricity and water so vast it threatens to overwhelm the country's plans for a renewable energy future. More than fifteen data centres are under construction right now, most built to feed the computational appetite of AI systems, adding to the 120 facilities already operating — one of Europe's densest concentrations per capita.
The numbers are unforgiving. Swiss data centres currently consume between 6 and 8 percent of national electricity, a figure experts project could reach 15 percent by 2030 — exceeding the government's own targets for expanding renewable capacity. The infrastructure powering AI's growth is outpacing the infrastructure needed to power it cleanly.
Swiss citizens have begun to notice. A survey by AlgorithmWatch CH across five European countries found four in five Swiss respondents worried about rising energy and water consumption from data centres. Nearly three-quarters believe new facilities should run entirely on renewable power. Asked to rank sectors by priority in an energy shortage, respondents placed data centres last — below healthcare, food supply, and everything else.
Water adds a second dimension of concern. Seven in ten respondents fear data centre cooling demands will damage local ecosystems. The public has grasped what policymakers have been slow to state plainly: AI infrastructure has a footprint, and that footprint is growing.
Yet transparency remains elusive. Angela Müller of AlgorithmWatch CH puts it directly — AI does not materialise from nowhere. It requires labour, energy, and water that must be accounted for. Around eight in ten survey respondents support mandatory disclosure of energy consumption, sources, and environmental impact. Two-thirds want companies to reveal which data centres they use and why. A similar share believes operators should contribute more than households toward grid investment costs.
AlgorithmWatch CH has proposed a clear framework: new data centres should be powered by additional renewable capacity rather than drawing from existing supply; transparency must extend through supply chains; and companies should be legally required to demonstrate measurable reductions in power and water use. These are not radical demands — they are the baseline expectations of a public that sees no contradiction between progress and environmental protection.
The Federal Council has committed to new AI regulations by end-2026, with growing parliamentary pressure to embed environmental safeguards into that strategy. The question is whether those rules will arrive in time — and whether they will bend the trajectory of data centre growth toward sustainability, or simply accommodate it.
Switzerland is facing a quiet crisis that few people see coming. Beneath the gleaming promise of artificial intelligence lies a hunger for electricity and water so vast that it threatens to overwhelm the country's carefully laid plans for a renewable energy future. More than fifteen data centres are under construction across the nation right now, most of them built to feed the relentless computational appetite of AI systems like ChatGPT. With over 120 facilities already humming in operation, Switzerland now hosts one of Europe's densest concentrations of data centres per capita—a distinction that comes with a cost the public is only beginning to grasp.
The numbers tell a stark story. Swiss data centres currently consume between 6 and 8 percent of the country's total electricity supply. That figure alone is substantial. But experts project it could nearly double to 15 percent by 2030—a threshold that would exceed the government's own targets for expanding renewable generation capacity. The math is unforgiving: the infrastructure required to power AI's growth is outpacing the infrastructure required to power it cleanly. Large-scale AI systems demand enormous amounts of computing power just to function, and that demand is accelerating.
What makes this moment significant is that Swiss citizens have begun to notice, and they are not pleased. A survey commissioned by AlgorithmWatch CH and conducted across five European countries reveals the depth of public concern. Four out of five Swiss respondents worry about rising energy and water consumption from data centres. Nearly three-quarters believe new facilities should operate entirely on renewable power. When asked to rank nine different sectors in terms of priority during a hypothetical energy shortage, Swiss citizens placed data centres dead last—below healthcare, food supply, and every other option presented. The message was clear: this is not a priority worth sacrificing for.
Water emerges as a second, equally troubling dimension. Seven in ten survey respondents fear that data centre water consumption could damage local ecosystems. The concern is not abstract. Data centres require enormous quantities of water for cooling, and in a country where water resources are finite and ecosystems are delicate, that demand carries real consequences. The public has connected the dots: AI infrastructure is not a victimless technology. It has a footprint, and that footprint is growing.
Yet transparency remains elusive. Some technology companies have begun disclosing their power and water consumption figures, but independent verification is rare and inconsistent. Angela Müller, managing director of AlgorithmWatch CH, frames the problem plainly: AI does not materialize from nowhere. It requires vast quantities of human labour, energy, and water—resources that must come from somewhere and be accounted for. The International Energy Agency and the International Telecommunication Union have both documented how AI is already inflating the energy bills and carbon emissions of major technology firms. The public wants to know what is happening, and it wants assurance that someone is watching.
The survey reveals what amounts to a mandate for action. Roughly eight in ten respondents support mandatory disclosure of current and projected energy consumption, energy sources, and environmental impact assessments. Two-thirds want AI companies to reveal which data centres they use and for what purpose. A similar proportion believes data centre operators should pay more than households and small businesses toward grid investment costs. The public is not asking for AI to disappear. It is asking for accountability.
AlgorithmWatch CH has laid out a framework for what that accountability might look like. Data centres should be powered by additional renewable energy capacity rather than siphoning existing renewable power away from other sectors. Transparency requirements must extend throughout supply chains. Companies should be legally obliged to demonstrate concrete, measurable efforts to reduce both power and water consumption. These are not radical demands. They are the baseline expectations of a public that understands progress and environmental protection are not opposing forces—they are conditions that must advance together.
The political machinery is beginning to move. Several initiatives in the Swiss parliament reflect growing interest in the issue. The Federal Council has committed to unveiling new AI regulations by the end of 2026, and there is mounting pressure to embed environmental safeguards into that strategy. The question now is whether those regulations will arrive in time, and whether they will be stringent enough to bend the trajectory of data centre growth toward sustainability rather than simply accommodating it. Switzerland stands at a crossroads: it can shape how AI infrastructure develops in its territory, or it can watch as that development shapes the country instead.
Citações Notáveis
AI does not fall from the sky—or from a 'cloud.' It requires vast amounts of human labour, energy and water.— Angela Müller, managing director of AlgorithmWatch CH
A forward-looking AI strategy must also confront its energy and water consumption. Progress and climate protection should go hand in hand.— Angela Müller, AlgorithmWatch CH
A Conversa do Hearth Outra perspectiva sobre a história
Why does Switzerland matter here? It's one country among many dealing with data centres.
Because Switzerland is a test case. It has the highest concentration of data centres per person in Europe, and its citizens are paying attention in a way that creates political pressure. What happens here could set a precedent for how other countries regulate the environmental cost of AI.
The survey shows 80% support for mandatory disclosure. That's overwhelming. Why hasn't that translated into law yet?
Because the technology is moving faster than regulation. Companies have no legal obligation to disclose these figures, and many prefer opacity. The Federal Council is still drafting AI rules—they won't arrive until 2026. By then, fifteen more data centres will be built.
You mentioned water as a second concern. Is that actually as serious as energy consumption?
In different ways. Energy is about climate and grid capacity. Water is about local ecosystems and scarcity. In a drought year, a data centre cooling system can strain aquifers and rivers. The public understands this intuitively—they ranked data centres last in a resource-shortage scenario.
What does AlgorithmWatch CH actually want to happen?
They want data centres powered by new renewable capacity, not existing capacity diverted from hospitals or homes. They want binding transparency rules throughout supply chains. And they want companies legally required to prove they're reducing consumption, not just reporting it.
Is there any sign the tech companies themselves are moving on this?
Some have started disclosing figures. But without independent verification or legal requirements, disclosure is voluntary and selective. Companies release what makes them look good. That's not accountability.
What's the real deadline here?
2030. That's when data centre electricity consumption could hit 15 percent of Switzerland's total supply—exceeding the government's renewable expansion plans. If regulations don't arrive before then, the infrastructure will already be locked in.