The Machine That Could Save or Scorch the Planet

A landmark UN report warns that artificial intelligence is quietly devouring water, land, and energy at a dangerous pace. But activists and scientists say the same technology, if steered wisely, could be humanity's most powerful weapon against the climate crisis.

When Edith Namyalo looks at the sky above Kampala, she does not just see clouds. She sees data rainfall patterns shifting, dry seasons stretching longer, the quiet signals of a climate already in distress. As a climate activist working on the front lines in Uganda, she has spent years watching a crisis unfold in slow motion. But lately, she has been paying close attention to something new: a technology that could either help her read those signals better, or make the underlying crisis dramatically worse.

That technology is artificial intelligence.

"If AI is used the right way, it could be a plus for researchers and policy makers," Namyalo told this reporter. She paused, then offered an analogy that felt both simple and precise: "AI is a growing child. If groomed well, we all can benefit greatly."

Her words capture a tension now at the centre of global climate debate, one that a sweeping new report from the United Nations has brought into sharp focus.

[Photo: Photo by İsmail Enes Ayhan on Unsplash]

THE UN’S ALARMING FINDINGS

Published on 4 June 2026 to coincide with World Environment Day, the UN University (UNU) report AI's Environmental Costs Threaten Water, Land and Climate delivers a warning that goes well beyond the familiar concern about carbon emissions.

The numbers are staggering. Data centres, the vast server farms that power AI services around the world, could consume 945 terawatt-hours of electricity every year by 2030. To put that in perspective, it is nearly triple the combined annual electricity use of Pakistan, Bangladesh, and Nigeria, three countries that are home to more than 650 million people. Here's more - 945 TWh/yr  projected data centre electricity use by 2030, 1.3 billion  people whose water needs could equal AI's water footprint, 14,500 km²  land footprint of AI infrastructure by 2030, twice the size of Jakarta and 2.5M tonnes of e-waste AI infrastructure may be generated annually by 2030

But electricity is only part of the story. Every server running an AI model must be cooled, and that cooling consumes enormous volumes of water. The UNU study found that AI-related water use could equal the basic annual domestic needs of 1.3 billion people by the end of the decade. Its land footprint, the physical space needed for power generation and supply chains, may exceed 14,500 square kilometres, roughly twice the size of the Jakarta metropolitan area.

Then there is the e-waste. AI infrastructure is projected to generate up to 2.5 million tonnes of discarded hardware annually by 2030. Much of that burden, the report warns, will fall on lower-income countries with limited capacity for safe disposal, nations that have contributed least to the AI boom.

NOT JUST TRAINING - YOUR DAILY PROMPT COUNTS

One finding in the UNU report is particularly striking: public debate has focused heavily on the energy cost of training large AI models, those initial, electricity-intensive runs that teach an AI system from scratch. But the study found that day-to-day usage accounts for roughly 80 to 90 per cent of total energy demand.

A single widely used AI service is estimated to process around 2.5 billion prompts per day, consuming hundreds of gigawatt-hours of electricity each year. And not all requests are equal. Generating a single AI image can require more than a thousand times the energy of a simple text task. Video generation is even more demanding, and both are growing in popularity at an extraordinary speed.

"AI is a growing child. If groomed well, we all can benefit greatly," said Edith Namyalo, who is a climate activist in Uganda.

A TECHNOLOGY THAT CONCENTRATES POWER AND COSTS

The UN report also raises uncomfortable questions about equity. More than 90 per cent of AI-specialized computing capacity is concentrated in just two countries: the United States and China. Meanwhile, over 150 nations lack significant domestic AI infrastructure.

This means the environmental costs, strained water supplies, overburdened power grids, and mountains of e-waste are often borne by communities that see little of the economic benefit. Expanding data centre facilities are, in some cases, drawing heavily on water supplies in regions already experiencing drought conditions.

"The benefits of the technology are global, but its costs are often concentrated in specific regions," the report states plainly. For activists like Namyalo, working in a country where climate impacts are already severe, this framing resonates deeply.

BUT THE SAME TOOL COULD HELP FIGHT BACK

The UN report is not a condemnation of AI. Its authors are careful to stress that the findings are not an argument against technology, but a call for it to be developed responsibly, within planetary limits. And there is a compelling counter-argument.

Research published in the journal npj Climate Action in March 2026 estimated that AI applications in power, food, and mobility systems could cut global greenhouse gas emissions by five to ten per cent by 2030 — equivalent to between 2.6 and 5.3 billion tonnes of CO₂. That is a meaningful chunk of what the world needs to shave off to stay within safe climate limits.

The ways AI could help are numerous. It can optimise power grids to absorb more renewable energy, reducing the need for fossil-fuel backup. It can improve weather and climate forecasting, giving governments better tools to prepare for floods, heatwaves, and droughts. It can help farmers use water and fertilisers more precisely, cutting the emissions that come from agriculture. It can detect methane leaks from oil and gas infrastructure using satellite imagery — leaks that would otherwise go unnoticed for months.

"AI can identify useful information within large amounts of unstructured data," notes UNESCO's Global AI Ethics and Governance Observatory. "It can analyze satellite imagery to monitor greenhouse gas emissions, pinpoint deforestation, or identify areas vulnerable to coastal flooding."

For researchers and policy makers in countries like Uganda, such tools could be transformative, providing access to high-quality climate data and analysis that was previously available only to well-resourced institutions in the Global North.

THE REBOUND PROBLEM

There is a catch, however. The UNU report highlights what economists call the rebound effect: as AI becomes cheaper and more efficient, its use grows rapidly, and total resource consumption rises even as individual operations get leaner.

Solutions that seem green in one dimension may create new pressures in others. Switching from fossil fuels to certain renewable energy sources may reduce carbon emissions, but can significantly increase water consumption and land use. The picture is rarely simple.

The UN has urged AI developers to run their data centres on 100 per cent renewable energy by 2030. But current reality falls well short: fossil fuels, primarily natural gas and coal, still supply the majority of data centre energy worldwide.

WHAT RESPONSIBLE AI LOOKS LIKE

The UN study outlines a framework it calls a "responsible AI ecosystem," built on principles of transparency, efficiency by design, equity, lifecycle responsibility, global cooperation, and sustainable use. Governments are urged to integrate AI infrastructure planning into their energy, water, and land-use strategies. Companies are encouraged to design systems that minimize resource consumption from the outset. And users, the report suggests, have a role to play too, choosing lower-impact applications where possible.

Namyalo's framing of AI as a growing child may be the most intuitive way to understand what is at stake. Children are not inherently good or bad. They become what they are shaped to be by the values instilled in them, the environment they grow up in, and the choices made by those who raise them.

"If AI is used the right way," she said, "it could be a plus for researchers and policy makers."

The urgent, unresolved, and increasingly consequential question by the day is whether the world's governments, technology companies, and communities will make the choices needed to groom it well.

Martha Nimusiima is a journalist, writer, and reporter with a passion for storytelling and a keen eye for detail.

Sources: 

UN University Report (June 4, 2026): AI's Environmental Costs Threaten Water, Land, and Climate. Full report: news.un.org/en/story/2026/06/1167658

Additional sources:

npj Climate Action (March 2026)

UNEP

UNESCO AI Ethics Observatory

World Economic Forum