The Hidden Environmental Cost of AI Is Bigger Than You Think
News Desk
Islamabad: As artificial intelligence (AI) transforms industries and becomes an integral part of daily life, a new United Nations report warns that its rapid growth could place unprecedented pressure on the world’s electricity, water and land resources by the end of the decade.
The report, released by the UN University Institute for Water, Environment and Health (UNU-INWEH), argues that AI’s environmental impact has been “systematically mismeasured” because public discussion has focused largely on carbon emissions while overlooking the technology’s growing demand for water, land and electricity.
According to the report, data centres powering AI are projected to consume 945 terawatt-hours (TWh) of electricity annually by 2030, more than double current levels. If data centres were a country, they would rank among the world’s largest electricity consumers.
Researchers estimate that global data centres consumed around 448 TWh of electricity in 2025, making them the equivalent of the world’s 11th-largest electricity user.
Beyond electricity, the report highlights an equally alarming rise in water consumption.
By 2030, AI-related data centres could require 9.3 trillion litres of water each year, enough to meet the basic annual domestic water needs of approximately 1.3 billion people in Sub-Saharan Africa.
The land footprint is also expected to expand significantly. By the end of the decade, AI infrastructure could occupy more than 14,500 square kilometres, an area roughly twice the size of the Jakarta metropolitan region.
The report cautions that reducing carbon emissions alone does not guarantee environmental sustainability.
“Low-carbon does not necessarily mean low-water or low-land,” the report noted, warning that some clean energy transitions can unintentionally increase pressure on freshwater supplies and land resources.
Kaveh Madani, Director of UNU-INWEH, stressed that the findings should not be interpreted as opposition to artificial intelligence.
“This report is not a case against artificial intelligence,” he said, urging governments, technology companies and researchers to promote responsible AI development while addressing its unintended environmental consequences.
The study also challenges a common misconception about AI’s energy use.
While public attention often focuses on the enormous computing power needed to train advanced AI models, researchers found that 80 to 90 percent of AI’s total electricity consumption comes from inference, the continuous process of generating responses to users after deployment.
With billions of interactions taking place every day, even routine AI usage has become a major source of energy demand.
The report estimates that ChatGPT alone processes around 2.5 billion prompts daily, consuming approximately 383 gigawatt-hours (GWh) of electricity annually.
Energy consumption also varies dramatically depending on the task. A single AI-generated image can require about 1,450 times more energy than a basic text-classification task, while generating a short AI video may consume as much electricity as processing 200,000 spam classifications.
The report concludes that AI’s environmental footprint extends far beyond carbon emissions and calls for a more comprehensive approach to sustainability.
It recommends building a responsible AI ecosystem based on transparency, efficiency by design, environmental justice, lifecycle responsibility, global cooperation and the sustainable use of natural resources.
As AI adoption accelerates worldwide, the report argues that balancing technological innovation with environmental stewardship will be essential to ensure that future progress does not come at the expense of the planet’s most critical resources.
