UN Report Warns AI Could Soon Use 3% of World’s Electricity and More Water than We Need to Drink

Why It Matters

The forecast highlights AI’s potential to become a major energy and water consumer, threatening climate targets and amplifying the digital divide, making sustainability a critical component of future AI policy.

Key Takeaways

• AI could consume 3% of global electricity by 2030
• Data centre water demand may reach 9.3 trillion litres annually
• Efficiency gains risk higher total AI usage via Jevons paradox
• 90% of AI cloud capacity is concentrated in US and China
• Report urges mandatory environmental disclosures for AI models

Pulse Analysis

The United Nations' latest environmental assessment warns that artificial intelligence could soon account for roughly three percent of the planet's electricity consumption by 2030, a level comparable to the entire power use of Saudi Arabia today. The projection assumes a doubling of data‑centre demand as AI models become more capable and cheaper to run. While industry leaders tout efficiency improvements, the report invokes the Jevons paradox – the tendency for lower operating costs to spur greater overall usage – suggesting that smarter algorithms may paradoxically amplify, rather than curb, energy draw.

Beyond power, the AI boom threatens water and land resources at an unprecedented scale. The UN study estimates that cooling data centres will require about 9.3 trillion litres of water each year, enough to meet the global population's drinking needs, and will occupy land roughly ten times the size of Mexico City. Carbon emissions from this expanded footprint would necessitate planting 6.7 billion trees over a decade to achieve neutrality. Moreover, the infrastructure is highly concentrated: 32 nations host AI‑specific clouds, with 90 percent of capacity located in the United States and China, widening the digital divide and shifting environmental burdens to resource‑rich but low‑tech regions.

Policymakers are urged to embed sustainability into the AI lifecycle. The UN roadmap calls for transparent environmental disclosures at both model and task levels, design‑by‑efficiency standards, and a governance framework that tracks mineral sourcing, recycling, and end‑of‑life disposal. Nations such as New Zealand and Australia have launched AI strategies, yet they lack mandatory reporting on energy or emissions, leaving a regulatory gap. Aligning AI development with climate goals will require coordinated global standards, incentives for low‑impact architectures, and a shift from a growth‑only mindset to one that balances innovation with planetary health.