UN Calls For AI Firms To Reveal Their Environmental Costs

The United Nations is urging AI companies to disclose the full environmental cost of their technologies, arguing that the industry’s impact extends far beyond electricity consumption and carbon emissions.

Why The UN Is Concerned

The warning follows the publication of a new report by the United Nations University Institute for Water, Environment and Health (UNU-INWEH), which argues that AI’s environmental footprint is being measured too narrowly.

While public debate has largely focused on carbon emissions and electricity consumption, the report says AI also has substantial water and land footprints that are often overlooked. Cooling data centres, generating electricity and building the infrastructure needed to support AI all consume natural resources on a scale that is growing rapidly.

The report concludes that greater transparency is urgently needed so governments, investors and the public can properly assess AI’s true environmental impact.

What The Numbers Show

According to the report, global data centres could consume around 945 terawatt-hours of electricity every year by 2030. That is almost three times the combined annual electricity consumption of Pakistan, Bangladesh and Nigeria, countries with a combined population of more than 650 million people.

Electricity, however, is only part of the story. The researchers estimate that AI-related electricity generation and cooling could require around 9.3 trillion litres of water each year by the end of the decade, equivalent to the basic annual domestic water needs of approximately 1.3 billion people in Sub-Saharan Africa.

The associated land footprint could exceed 14,500 square kilometres, roughly twice the size of the Jakarta metropolitan area.

Professor Kaveh Madani, Director of UNU-INWEH and one of the report’s authors, said: “This report is not a case against artificial intelligence… It is a call for using it responsibly and addressing its unintended impacts proactively to make it sustainable and equitable.”

Looking Beyond Carbon

One of the report’s central arguments is that measuring AI sustainability through carbon emissions alone can be misleading.

Researchers found that different energy sources can have very different environmental consequences. For example, an energy source that produces fewer greenhouse gas emissions may require considerably more water or land.

Lead author of the report, Dr Miriam Aczel, said: “If we keep judging AI sustainability by carbon alone, we might think that renewables make AI infrastructure clean, but that is solving one problem while creating other problems, often in places that didn’t ask for it.”

The report therefore calls for AI companies to publish standardised information covering carbon, water and land use together, allowing meaningful comparisons between different technologies and data centres.

Daily AI Use Is Driving Demand

The report also challenges another widely held assumption. Many people associate AI’s environmental impact with training large language models. However, the researchers estimate that running AI systems after they have been deployed, known as inference, now accounts for around 80 to 90 per cent of total AI energy consumption.

As millions of people generate text, images and videos every day, those routine interactions collectively consume far more electricity than the original training process.

The report also highlights how energy requirements vary dramatically depending on the task being performed. Generating an AI image can require around 1,450 times more energy than a simple text classification task, while AI video generation demands considerably more still.

A Wider Sustainability Challenge

The United Nations also highlights broader environmental issues extending beyond electricity use. For example, by 2030, AI infrastructure could generate up to 2.5 million tonnes of electronic waste every year, while demand for critical minerals used in processors, batteries and other hardware continues to increase.

The report argues that the environmental burdens associated with AI are often concentrated in communities hosting data centres, mining operations and electronic waste processing, while many of the economic benefits flow elsewhere.

UN Under-Secretary-General and United Nations University Rector Professor Tshilidzi Marwala said: “AI can certainly advance prosperity and human well-being. Whether it does so equitably is now a governance question, not a technical one.”

Towards More Responsible AI

Rather than arguing against AI, the report proposes what it describes as a “responsible AI ecosystem” built around six principles: transparency, efficiency by design, equity, lifecycle responsibility, global cooperation and sustainable use.

Governments are encouraged to incorporate AI infrastructure into energy, water and land-use planning, while AI developers are urged to improve efficiency and publish consistent environmental reporting.

The report also suggests organisations deploying AI should consider selecting the least resource-intensive model capable of completing a particular task rather than automatically using the largest available systems.

What Does This Mean For Your Organisation?

For organisations adopting AI, the UN’s report highlights how sustainability is becoming an increasingly important part of AI governance rather than simply a data centre issue.

Many businesses are already assessing suppliers on environmental, social and governance criteria. As AI becomes embedded across more business applications, organisations may increasingly expect technology providers to disclose not only carbon emissions but also water use, land impacts and other environmental costs associated with their AI services.

It now looks as though transparency is likely to become just as important as technical performance. As businesses invest more heavily in AI, questions about how those systems are powered, where they operate and what resources they consume are likely to become an increasingly important part of procurement, sustainability reporting and corporate responsibility. The United Nations is making it clear that measuring AI’s environmental impact should no longer stop at carbon emissions alone.