UN report: AI could double electricity use by 2030, drain global water supply
Synopsis
The United Nations is sounding a stark alarm: AI's electricity use could double by 2030, its carbon footprint could rival the UK's, and its water demand could outstrip what the entire world drinks in a year. The culprit isn't just growth — it's the Jevons paradox, where efficiency gains make AI cheaper and trigger even greater consumption. A resource reckoning may be coming faster than the industry expects.
Key Takeaways
A UN report warns AI could double its electricity consumption by 2030 , reaching roughly 3 per cent of global power use.
AI's greenhouse gas emissions could equal those of the United Kingdom ; offsetting the carbon footprint would require 6.7 billion trees grown over ten years.
Data centres already consume as much electricity as Saudi Arabia annually and may need 9.3 trillion litres of water to support projected growth.
The Jevons paradox means AI efficiency gains are likely to drive higher, not lower, total resource consumption.
Only 32 nations host AI cloud infrastructure; 90 per cent of capacity sits in the US and China , leaving other countries to absorb environmental costs.
The UN calls for routine environmental disclosures and full value-chain governance from mineral sourcing to e-waste disposal.
A new United Nations report has warned that artificial intelligence could double its electricity consumption by 2030, accounting for roughly 3 per cent of global power use and generating greenhouse gas emissions comparable to those of the United Kingdom. The report, released around 5 June, also cautioned that AI's water demand for cooling could exceed the annual drinking water needs of the entire global population.
The Scale of AI's Energy Footprint
According to the UN report, data centres already consumed as much electricity as Saudi Arabia last year. Should electricity use double by 2030, the resulting carbon footprint would require 6.7 billion trees grown over a decade to offset. Beyond energy, data centres are projected to need approximately 9.3 trillion litres of water and occupy land nearly ten times the size of Mexico City to sustain that level of growth.
The Jevons Paradox and AI Efficiency
The report invokes the concept of the Jevons paradox — first observed by economist William Stanley Jevons in 19th-century England, where efficiency gains in coal use lowered costs, ultimately driving higher overall consumption rather than reducing it. The UN warns that AI faces the same dynamic: as models become cheaper and more capable, new use cases and higher usage volumes will emerge, eroding any savings from efficiency improvements. In effect, making AI more efficient may simply accelerate its adoption and total resource draw.
A Widening Digital and Environmental Divide
The report highlights a stark geographic imbalance: only 32 nations currently host AI-specific cloud infrastructure, with 90 per cent of that capacity concentrated in the United States and China. Countries that consume AI services without hosting the infrastructure bear a disproportionate environmental burden through mineral extraction and e-waste. This emerging divide, the report argues, risks compounding existing inequalities between the Global North and South.
What the UN Is Recommending
To avoid what it describes as an unsustainable trajectory, the UN has laid out a roadmap for responsible AI development built on six guiding principles: transparency, efficiency by design, equity and justice, lifecycle responsibility, global cooperation, and sustainable use. Critically, the report urges routine environmental disclosures at both the model and task level, and calls for full value-chain governance — from mineral sourcing through to recycling and safe disposal of hardware.
What Comes Next
The report's findings arrive as AI investment continues to accelerate globally, with major technology companies announcing ever-larger data centre buildouts. Whether voluntary disclosure frameworks will be sufficient — or whether binding international standards will be required — remains an open question. The UN's call for global cooperation suggests that unilateral national policies may be inadequate to address a problem that is, by its nature, borderless.
Point of View
Greener chips, smarter inference. The Jevons paradox argument cuts through that framing cleanly: cheaper AI is not greener AI, it is simply more AI. What the report does not fully resolve is enforcement — voluntary disclosure frameworks have a poor track record in extractive industries, and there is little reason to expect tech to be different. The geographic concentration of infrastructure in the US and China also means the two countries most resistant to binding multilateral frameworks are the ones whose cooperation is most essential. Without hard targets and independent verification, this report risks becoming another well-cited warning that changes nothing.
NationPress
21 Jul 2026
Frequently Asked Questions
What does the UN report say about AI and electricity use by 2030?
The UN report warns that AI could double its electricity consumption by 2030, reaching approximately 3 per cent of global power use. The associated carbon footprint would be comparable to the greenhouse gas emissions of the United Kingdom.
How much water could AI consume, according to the UN?
According to the report, data centres supporting AI growth could require around 9.3 trillion litres of water — more than the annual drinking water needs of the entire global population. This water is primarily used for cooling data centre infrastructure.
What is the Jevons paradox, and why does it apply to AI?
The Jevons paradox is the observation that efficiency improvements in resource use tend to increase total consumption rather than reduce it, because lower costs drive greater adoption. The UN report argues that as AI models become cheaper and more capable, new use cases will multiply, erasing any environmental savings from efficiency gains.
Which countries host most of the world's AI infrastructure?
Only 32 nations currently host AI-specific cloud infrastructure, and 90 per cent of that capacity is concentrated in the United States and China. Countries that use AI services without hosting infrastructure bear a disproportionate share of the environmental costs through mineral extraction and e-waste.
What solutions does the UN recommend to reduce AI's environmental impact?
The UN recommends a roadmap built on six principles: transparency, efficiency by design, equity and justice, lifecycle responsibility, global cooperation, and sustainable use. It specifically calls for routine environmental disclosures at both the model and task level, and full value-chain governance from mineral sourcing to hardware recycling and disposal.
