The United Nations has issued a stark warning, urging countries to confront the escalating environmental costs of artificial intelligence (AI) as its rapid expansion intensifies demands on energy, water, and carbon emissions.
AI data centres are notoriously resource-intensive, consuming vast quantities of electricity, water for cooling, and significant land.
A report from the Canada-based UN University Institute for Water, Environment and Health (UNU-INWEH) also highlighted the technology's footprint from chip production, critical minerals, and electronic waste.
Beyond physical infrastructure, the cumulative impact of billions of user interactions – every prompt, generated image, and query – significantly contributes to this burden.
The report implored governments, businesses, and investors to embed environmental considerations at the core of all AI decisions, ensuring its development remains sustainable.
It comes as the global AI market is predicted to grow 25-fold to nearly five trillion dollars by 2033.
“When a technology scales this quickly, unintended social, economic, geopolitical and environmental consequences can accumulate quietly, and then become complex and inequitable to correct once systems, investments and dependencies lock in,” the report said.
“Public debate has rightly focused on AI risks such as bias, privacy, disinformation, labour disruption and inequity.
“Yet one of the most consequential dimensions of AI that remains comparatively underexamined is its environmental footprint and the justice implications that follow from where and how AI infrastructure expands.”
The AI boom is already driving soaring energy consumption, with AI-related workloads accounting for roughly 20% of total data centre electricity use in 2025, according to the scientists.
This share is expected to rise to 40% by the end of the decade, they said.
This means AI-linked electricity consumption is set to increase to such a level by 2030 that it could meet the residential electricity needs of all 1.3 billion people living in sub-Saharan Africa for more than two years, according to the findings.
Current forecast trajectories also suggest total electricity demand from data centres – the physical backbone of AI – could roughly double by 2030 as the sector grows.
Producing that much electricity could have a carbon footprint of nearly 400 million tonnes of planet-heating carbon dioxide, the report said, requiring 6.7 billion trees grown over a decade to offset – about twice the number of trees in the UK.
The estimated 9.3 trillion litres of water used by data centres by 2030 would meet the drinking water needs of Earth’s 8.1 billion people for about 1.6 years, the report said.
AI infrastructure is also predicted to generate up to 2.5 million tonnes of e-waste each year by 2030.
The UN experts challenged the assumption that renewable-powered data centres are always green or sustainable – a finding that cuts against a lot of industry messaging.
If a data centre is powered by “low carbon” electricity generation, this does not necessarily make the AI generation less impactful because it could still be linked to significant water or land use, the paper said.
The scientists also identified a widening digital divide between nations that build and control AI systems, and those that only consume them while bearing a disproportionate share of the environmental costs.
Only 32 nations host AI-specialised cloud infrastructure and 90% of that capacity is in the US and China, they said.
More than 150 countries have no sovereign AI computing at all, according to the report, with excluded countries often facing mineral extraction and e-waste burdens while the strategic benefits flow elsewhere.
To ensure the technology grows in a sustainable way, governments are being urged to ensure decisions on AI infrastructure are integrated with energy planning, water governance and land-use permitting.
Industry and developers should consider environmental impacts in model selection, default outputs and routing decisions, and improve efficiency by design, the report said.
Meanwhile, users should choose the lightest model and lowest-energy format that meets the task they are using the technology for while data centre operators should approach decisions around locations and procuring energy as an environmental one.
And investors should treat electricity, carbon, water and land footprints as material risks while communities and civil society should be involved early in decisions around locations for data centres.
Tshilidzi Marwala, UN under-secretary-general and a co-author of the report, said: “The promise of AI is immense, particularly in areas such as healthcare, education, scientific discovery and climate resilience.
“But innovation without stewardship risks deepening inequality and intensifying pressure on already stressed planetary systems.”
