AI to chew through land, water and power by 2030, UN warns
By Dr Joe Milton, the Australian Science Media Centre
Artificial intelligence (AI) could consume enough electricity by 2030 to power all of Sub-Saharan Africa for more than five years, while placing growing pressure on water supplies, land use and waste management, according to a new report.
The report, from the United Nations University Institute for Water, Environment and Health, says AI’s environmental footprint extends well beyond carbon emissions, and warns that the focus on these risks missing the impacts of the technology's energy, water and land use.
Professor Niusha Shafiabady, Head of Information Technology (IT) at the Australian Catholic University, told the AusSMC the findings are a reminder that we often overlook the many hidden real-world environmental costs of clever computing.
"Artificial intelligence may feel weightless and virtual, but this report makes clear that it is firmly rooted in the physical world," she said.
"Every prompt, every model and every data centre draws on energy, water, land and minerals, and the scale of that demand is growing as AI adoption accelerates."
But, she added that AI should be considered in context.
"AI is not the largest environmental threat we face. The far greater risks come from fossil fuel emissions, land degradation, industrial agriculture and global supply chains."
Associate Professor Walayat Hussain, who leads the AI for Decision Excellence (AIDX) Lab at the Australian Catholic University, said AI is part of a broader digital infrastructure already consuming significant resources.
"Data centres, cloud services, social media, video streaming, e-commerce, online banking and large-scale digital storage were already consuming electricity, water, land and cooling resources long before the recent rise of generative AI," he said.
"AI is adding new demand, and in some areas accelerating it sharply, but it is not scientifically accurate to place the whole burden of the digital infrastructure problem on AI alone."
The report estimates AI-related electricity use could reach 945 terawatt-hours annually by 2030 - or nearly 3% of global demand - with emissions of up to 400 million tonnes of carbon dioxide.
It also estimates the land required for AI data centres could exceed 14,000 square kilometres, roughly the size of Northern Ireland, while annual water use could reach 9.3 trillion litres. AI infrastructure could also generate up to 2.5 million tonnes of electronic waste each year by the end of the decade.
Professor Geoff Webb, Australian Laureate Fellow in data science and AI at Monash University, said the report highlights environmental issues that must be considered alongside discussions of the social and economic impacts of AI.
"I welcome this comprehensive report on AI’s significant environmental impacts," he said.
"As if they are not enough, these concerns should be considered alongside the major social and economic effects that the report mentions only briefly and does not address in its action plan."
He said tax breaks and other factors are making it easier for businesses to adopt AI, and warned the technology could have wider effects on jobs, society and culture.
The report notes that while training AI is energy intensive, the combined impact of billions of daily searches may be greater, because searching the web using AI uses around 10 times as much electricity as non-AI searches.
Professor Daswin De Silva, Co-Director of the La Trobe AI Institute, said the report provides a useful assessment of AI’s environmental footprint, while noting that the technology is already becoming more efficient.
The report calls for coordinated action across governments, industry and users to improve transparency, efficiency and environmental stewardship.
"We must act now to address these profound issues if we are to prevent extraordinary environmental degradation, social disruption and cultural decline," said Prof Webb.
A/Prof Hussain said the solution is not to reject AI outright, but to govern it more carefully.
"The right debate is not ‘AI is good’ or ‘AI is bad’," he said. "It is about responsible use. We should reduce wasteful, high-energy applications and demand transparency from industry, while protecting high-value uses that expand access to education, healthcare and opportunity. If governed carefully, AI could help narrow some inequalities rather than simply deepen them."
You can read the Expert Reaction in full here
This article originally appeared in Science Deadline, a weekly newsletter from the AusSMC. You are free to republish this story, in full, with appropriate credit.
Contact: Dr Joe Milton
Phone: +61 8 7120 8666
Email: info@smc.org.au