The Environmental Cost of AI: A Growing Challenge for Sustainability
Artificial intelligence (AI) is reshaping industries and economies, but its environmental impact is raising urgent concerns. A United Nations University Institute for Water, Environment and Health (UNU-INWEH) report highlights that AI’s energy consumption could double by 2030, consuming 3% of global electricity and straining water and land resources. This projection underscores the need for a reevaluation of AI development and usage to align with sustainability goals.
AI’s Energy Demand and Environmental Footprint
The UNU-INWEH report, published in 2026, emphasizes that AI’s environmental costs depend not only on energy volume but also on the source of that energy. For instance, low-carbon electricity does not necessarily translate to low water or land use. By 2030, AI’s energy use could reach 3% of global electricity, with associated carbon emissions equivalent to the UK’s annual output. Additionally, the report estimates that AI systems would require 9.3 trillion liters of water for cooling—more than the annual drinking water needs of the global population.
These figures align with the International Energy Agency’s (IEA) forecast that data center electricity consumption will more than double to 945 TWh by 2030. In 2025, data centers already consumed as much electricity as Saudi Arabia, the 11th-largest electricity consumer globally. If AI demand grows as projected, the carbon footprint would require 6.7 billion trees to offset over a decade.
The Jevons Paradox and AI Efficiency
Efficiency gains in AI development may not reduce overall resource use, according to the Jevons paradox. This economic principle, named after 19th-century economist William Stanley Jevons, posits that improved efficiency leads to increased consumption rather than reduced demand. The UNU-IN