AI may soon consume water equaling consumption of 1.3bn people:UN warns, by 2030 data centres can use more electricity than Pak, Bangladesh, Nigeria combined

Artificial intelligence (AI) is becoming a major part of everyday life. From writing emails and creating images to helping students learn and professionals work faster, AI is making things easier than ever. But behind this convenience, scientists are raising an important concern: Is AI putting too much pressure on Earth’s resources? A new report from the United Nations University (UNU) warns that AI’s rapid growth could bring serious environmental problems, especially because of its huge demand for electricity, water, and land. The hidden environmental cost of AI Most people think of AI as digital and invisible. But every AI tool runs on massive data centres, giant facilities filled with computers that store and process information. These data centres consume enormous amounts of power. According to the UNU report, by 2030, AI-powered data centres could use 945 terawatt-hours of electricity every year. That is nearly three times the combined annual electricity consumption of Pakistan, Bangladesh, and Nigeria, countries with more than 650 million people. And electricity is only part of the story. These centres also require huge amounts of water for cooling systems and energy production. Researchers estimate that their yearly water use could equal the basic domestic water needs of all 1.3 billion people living in Sub-Saharan Africa. The land needed to support AI infrastructure is also expanding quickly. By 2030, the total land footprint could exceed 14,500 square kilometres which is nearly twice the size of Jakarta’s metropolitan area, home to over 32 million people. Why everyday AI use is the bigger problem? Many people assume training advanced AI models consumes the most energy. Surprisingly, the report says the real issue is daily usage. Researchers found that 80–90% of AI’s total energy demand comes from everyday use, not training. Think about it: Millions even billions of people now use AI every day. One popular AI service reportedly handles around 2.5 billion prompts daily, using hundreds of gigawatt-hours of electricity each year. The type of task also matters. Simple text tasks use relatively less energy, but generating a single AI image can require over 1,000 times more energy than basic text classification. Video generation demands even more computing power. Although technology is becoming more efficient, experts warn this may not solve the problem. When AI becomes cheaper and faster, more people use it increasing total energy demand instead of reducing it. The environmental burden is not shared equally While the benefits of AI are global, its environmental costs are often concentrated in specific places. In some countries, data centres already consume a major share of national electricity supplies, putting pressure on energy systems. In others, these facilities are heavily using local water resources, sometimes even in drought-prone regions. Another growing issue is electronic waste (e-waste). By 2030, AI infrastructure could generate up to 2.5 million tonnes of e-waste every year. Lower-income countries may face the biggest challenge because many lack proper systems for safe disposal. The demand for minerals needed to build AI hardware is also increasing. Mining these critical materials can damage ecosystems and deepen social inequalities in extraction regions. A growing gap between countries The report also highlights a widening digital divide. Today, more than 90% of AI-specialised computing power is concentrated in just two countries – the United States and China. Meanwhile, over 150 countries have little or no major AI infrastructure. This imbalance creates difficult questions. Some nations may bear the environmental damage caused by AI infrastructure without fully benefiting from the economic growth AI can bring. Have humans created another unintended problem? History shows that many inventions solve one problem while creating another. Cars made travel easier but increased pollution. Plastic improved convenience but created a waste crisis. Social media connected people but also sparked concerns around misinformation, privacy, and mental health. Could AI follow the same pattern? According to Scientific American, data centres already account for around 1.5% of global electricity use, and demand is expected to rise sharply as AI adoption grows. Can AI become more responsible? Despite the alarming findings, researchers are not against AI. Instead, they say urgent action is needed to make sure AI grows in a sustainable way. The UNU report proposes a “responsible AI ecosystem” focused on transparency, efficiency, fairness, sustainable use, and global cooperation. Governments are being urged to include AI infrastructure in energy, water, and land planning. Companies are encouraged to build systems that use fewer resources, while users can help by choosing lower-impact AI tools whenever possible. The big question now is simple but urgent: Can the world enjoy AI’s benefits without paying an environmental price too high to afford?