Want to save the planet? Stop being so polite to AI chatbots
A UN-backed report says AI has a large hidden footprint in power, water, land and e-waste, and even polite chatbot prompts add up.
Intelligence analysis by GPT-5.4 Mini
A new UN report argues that AI’s environmental cost goes far beyond carbon. It says data centres already use vast amounts of electricity and water, and warns the footprint could grow sharply by 2030 as AI adoption rises.
AI is not just words on a screen. It needs giant computer buildings that use lots of electricity and water, like a thirsty factory that never sleeps. The article says even tiny chat habits can add up when millions of people use them.
Analysis
The hidden footprint
A new report from the Institute for Water, Environment and Health under the United Nations University says AI’s environmental cost is much broader than carbon emissions. The report points to electricity use, water consumption, land use and electronic waste as major parts of the problem.
The article says that data centres consumed an estimated 448 terawatt-hours of electricity last year. Put another way, if data centres were a country, they would rank 11th in the world for energy use, roughly alongside France. The report also says generating that electricity required about 4.5 trillion litres of water and 6,900 square kilometres of land.
Growth is expected to continue
The report projects AI-related electricity consumption could reach 945 TWh by 2030, or nearly 3 per cent of global electricity demand. It also warns that AI infrastructure could create up to 2.5 million tonnes of e-waste a year by then.
The piece also notes that the burden is unevenly distributed. According to the report, the United States and China host 90 per cent of the world’s AI-specialised cloud infrastructure, which means smaller countries have limited control over access, pricing and data governance.
The article uses a striking example to show how even small user habits can add up: skipping “please” and “thank you” in prompts could save enough energy to power the annual needs of 760,000 people in sub-Saharan Africa. The broader message is that AI’s growth carries real physical costs, and the report urges more sustainable practices as usage rises.
Key points
- A UN University-linked report says AI’s footprint includes electricity, water, land use and e-waste, not just carbon emissions.
- Data centres used an estimated 448 TWh of electricity last year, a level the report compares to the world’s 11th-largest energy user.
- The report says that electricity use required about 4.5 trillion litres of water and 6,900 sq km of land.
- AI-related electricity demand could reach 945 TWh by 2030, nearly 3 per cent of global power use.
- The report says the United States and China host 90 per cent of AI-specialised cloud infrastructure.
- It argues that even small prompt habits, like skipping politeness, can add up at scale.
If AI companies and users adopt more sustainable practices, the report suggests some of the environmental burden could be reduced. Better planning around energy, water use and waste could make AI growth less damaging while still allowing the technology to expand.
If AI usage keeps rising without changes, the report says electricity demand, water use and e-waste could all climb sharply by 2030. The concentration of AI infrastructure in the United States and China could also leave smaller countries with little control over access, pricing or data governance.
