A new United Nations environmental and sustainability report warns that by 2030 artificial intelligence (AI) could consume roughly 3 percent of global electricity production. The report also estimates that AI data centers and related systems will use about 9.3 trillion liters of fresh water per year — more than the world’s entire population drinks annually.
What drives the increase?
Technology companies frequently point out that improvements in algorithms and the introduction of new chips increase efficiency, lowering the energy required for individual tasks. The UN report cautions, however, that lower costs and greater availability often spur increased overall use. This is the Jevons paradox from economics: William Stanley Jevons observed in 19th-century England that efficiency gains in coal use led not to lower total consumption but to expanded use and higher aggregate demand.
Applied to AI, the effect means that even if single models or chips become more efficient, the overall energy and water footprint can still rise because use expands as the technology becomes cheaper and more attractive.
Carbon footprint and tree-planting example
Amanda Turnbull-McRae, a lecturer in law at the University of Waikato, wrote in The Conversation that the UN’s projected 3 percent share of global electricity would equate to a level of carbon dioxide emissions comparable to the United Kingdom’s annual output. Compensating for that carbon footprint would require planting and cultivating about 6.7 billion trees over ten years, by the cited estimate.
Water and energy demand, and geographic concentration
The UN report singles out the water consumption of data centers as particularly alarming: servers and their cooling systems require vast quantities of clean freshwater, and the report estimates annual water use at 9.3 trillion liters. The geographic concentration of AI infrastructure is also stark: AI-specific cloud infrastructure exists in only 32 countries, and roughly 90 percent of that capacity is located in the United States and China.
Such concentration deepens global technological and economic inequalities: many advanced economies and large corporations reap the benefits of AI, while developing countries frequently bear disproportionate environmental burdens — for example through mineral extraction and electronic waste.
Calls for transparency and regulation
The report stresses the need for government action: lawmakers should require tech companies to publish transparent data on energy and water use. At present, reliable public information on the true environmental costs of these technologies is often lacking, even as AI is being rapidly integrated across economic sectors.
Why this matters
AI undeniably offers major societal benefits: it can accelerate drug discovery, support climate action, and boost productivity. The UN report underlines, however, that without prompt and sustainable management of AI’s growing energy and water demands, and without addressing the geographic and economic inequalities around infrastructure and supply chains, social and environmental costs could intensify. The report therefore urges greater transparency and responsible regulation rather than unrestrained deployment.
