According to a new United Nations report, artificial intelligence could consume 40% of global data center electricity by 2030, representing a dramatic increase from current usage patterns. As reported by Associated Press, the report estimates that AI's energy demand would produce emissions equivalent to the UK's total output and deplete more water for cooling than the annual drinking water needs of the global population. The study, released by the United Nations University Institute for Water, Environment and Health (UNU-INWEH), warns that global data center electricity demand could exceed 945 terawatt-hours by 2030, representing roughly 3% of total global electricity usage. This amount of energy is comparable to what could power the entire population of 1.3 billion people in Sub-Saharan Africa for several years. Last year, data centres already consumed as much electricity as Saudi Arabia, which ranks as the world's 11th largest electricity consumer, with AI accounting for about one-fifth of that total. The report also notes that 90% of AI-related power use comes from operational requests rather than model training, including prompts, searches, and media-generation tasks. As reported by Associated Press, GPT alone accounts for 2.5 billion prompts a day, highlighting the operational intensity of current AI usage.
The UN report reveals that water consumption by data centers is projected to increase dramatically from 4.5 trillion liters in 2025 to 9.3 trillion liters by 2030. As reported by Tech Wire Asia, this water demand is linked to cooling systems and electricity use, with large-scale water withdrawals potentially straining aquifers and river systems, particularly in arid or groundwater-depleted regions. The report estimates that data center land use could expand from 6,900 square kilometers in 2025 to more than 14,500 square kilometers by 2030. The physical footprint of data centers is also expected to grow significantly as additional capacity is built. The report emphasizes that data center expansion could contribute to rising electronic waste and place additional pressure on land, power, and water resources if governments do not account for environmental costs. According to Associated Press, the study focused on energy use and didn't examine the massive amount of water used to cool data centers, though the water demand is linked to the enormous electricity consumption.
The UN report anticipates that AI will follow the Jevons paradox, an economic principle where technological improvements in efficiency lead to increased total consumption rather than decreased use. According to the report, as AI models become cheaper and more attractive, this is expected to encourage new uses and higher volumes of consumption, eroding any potential savings from efficiency advances. The paradox is named after economist William Stanley Jevons, who observed this effect with coal usage in 19th-century England, where efficiency gains did not reduce overall consumption but instead expanded use. The report notes that a typical ChatGPT-style query is about 200 times more energy-intensive than basic text classification, with image and video generation requiring even more computing power. Larger models require more electricity to train, and the report estimates that data centers produced 189 million tonnes of carbon dioxide emissions last year, with emissions expected to rise to 399 million tonnes by 2030. As reported by Associated Press, GPT-3 used about 1.3 billion watt-hours to train, but the next version used 50 to 70 billion watt-hours, demonstrating the exponential growth in energy requirements.
To avoid the Jevons paradox trap, the UN report lays out a roadmap for responsible AI use based on guiding principles of transparency, efficiency by design, equity and justice, lifecycle responsibility, global cooperation, and sustainable use. According to the report, responsible AI requires full value-chain governance from mineral sourcing to recycling and safe disposal, with the need for environmental disclosures to become routine in AI development. The report calls for incorporating projected AI demand in climate and energy planning, emphasizing the twinning of capability and environmental stewardship. The United Nations has urged policymakers to ensure sustainable AI development by mandating energy efficiency standards, transparency measures, and lifecycle management practices to reduce future resource pressure. The European Union is separately preparing minimum energy-efficiency standards for data centers, along with a possible sustainability label covering water use and clean energy supply. As reported by Associated Press, Cornell University energy engineering professor Fengqi You noted that the report's value lies in the U.N.'s credibility and authority, stating that "Its value is that a U.N. institution is putting carbon, water, land, life-cycle impacts and environmental justice into one frame" for an issue often shrouded in secrecy and partial disclosures.
The report highlights significant transparency challenges in the data center sector, with limited company disclosure making it difficult to assess where data centers are located, how large they are, and how much energy and water they consume. As reported by Associated Press, study co-author Miriam Aczel noted that "We cannot manage what companies do not disclose", while Cornell's Fengqi You emphasized that "Many companies and places are not transparent about what data centers and AI are consuming or even where and how big they are." The International Energy Agency has noted that data centers tend to concentrate in specific locations, creating grid integration challenges. Kaveh Madani, director of the institute and lead author of the report, emphasized that public debate often treats AI as software while overlooking the infrastructure behind it, stating that "AI is also physical infrastructure" including data centers, electricity generation, cooling systems, transmission networks, chips, minerals, land, and water. The report warns that poorly planned data center growth could add pressure in specific locations already facing resource constraints, while the natural environment should be at the center of thinking, calling for a rethink of the AI innovation playbook and shift toward a sustainable tech future.
