When you ask AI a question, the planet pays a price

Every time you type a prompt into an AI chatbot, it feels instant and effortless. But behind that quick reply is an energy-hungry infrastructure that consumes electricity, water, and, depending on where the data centre is located, significant amounts of fossil fuels. As the UN Environment Programme warns, the environmental footprint of artificial intelligence is expanding rapidly and remains largely invisible to the public.

The hidden energy and water cost of every AI prompt

AI systems rely on massive data centres packed with servers and high-performance chips. When you send a prompt, these machines perform complex computations that require continuous electricity. In many regions, most of this electricity still comes from fossil fuels, meaning every interaction with an AI tool indirectly emits greenhouse gases. Even a simple chatbot query typically uses more power than an ordinary web search, and when multiplied by millions of users, the cumulative impact becomes substantial.

Water is another critical but often overlooked part of the equation. The servers powering AI models generate intense heat that must be controlled through large-scale cooling systems, many of which depend on fresh water. The electricity that feeds these data centres also has a hidden water cost, since fossil-fuel and thermal power plants rely heavily on water for steam production and cooling. Even the hardware itself, from semiconductor fabrication to GPU manufacturing, requires vast amounts of ultrapure water. While a single AI prompt may consume only a few hundred millilitres of water when direct and indirect usage are combined, the total footprint becomes enormous at global scale. By 2027, AI-related data centres are expected to consume billions of cubic metres of water, often in regions already facing water stress.

AI for climate solutions… at what environmental price?

This growing concern over AI’s environmental cost is now entering global climate discussions. While artificial intelligence isn’t part of the formal negotiations at COP30, it is gaining visibility within the Action Agenda, a platform mobilizing voluntary climate action from civil society, businesses, investors, cities, and states. These actors may not sign treaties, but they are essential to transforming climate commitments into reality.

Brazil’s Government, for example, is mapping promising cases of how AI can strengthen climate resilience, a trend increasingly echoed across other UN Member States. One standout example comes from Lao People’s Democratic Republic, where researcher Alisa Luangrath created an AI-powered irrigation system in Savannakhet Province, a region suffering from severe water shortages and climate pressure. Her system integrates soil-moisture sensors, groundwater monitors, and meteorological data into an AI-driven analytical model capable of forecasting land conditions, water availability, and risks of flooding or extreme heat. Farmers receive real-time alerts through a mobile app, allowing them to plan planting and irrigation cycles more efficiently and reduce vulnerability to climate shocks.

But as AI’s potential for climate solutions grows, so do concerns about its environmental footprint. Luã Cruz, Coordinator for Telecommunications and Digital Rights at Brazil’s Consumer Defense Institute (Idec), warned that even everyday digital activities, from mobile phone use to routine online interactions, depend on massive data centres. These facilities, he noted, require enormous amounts of energy and water for cooling, occupy extensive land areas, and rely on mineral-intensive supply chains for their electronic components. Cruz added that many data centres “ignore planetary boundaries,” seeking locations with weak environmental regulation or generous tax exemptions, which further magnifies their ecological impact.

In a world where AI is becoming deeply embedded in daily life, it is crucial to recognize that digital convenience is not cost-free. Every prompt triggers a chain of resource consumption that places added pressure on the planet’s fragile ecosystems. The challenge now is to ensure that AI’s benefits, from climate-smart agriculture to advanced analytics, do not come at the expense of the environmental limits we cannot afford to cross.