A new United Nations report reveals how AI’s rapid growth is driving massive increases in electricity consumption, water use, land requirements, and environmental pressure around the world.
Key Takeaways
- A new United Nations report finds that AI’s environmental footprint extends far beyond electricity use, affecting carbon emissions, water resources, land use, and critical mineral supply chains.
- Global data center electricity consumption could exceed 945 TWh annually by 2030, enough to rank among the world’s largest electricity consumers.
- Training and operating advanced AI models requires enormous amounts of energy and water, raising important sustainability and environmental justice concerns.
- The findings reinforce a growing theme across Voices.Earth that digital technologies are not separate from the physical world and carry real environmental consequences.
Artificial intelligence is often portrayed as a digital revolution. However, what receives far less attention is the physical infrastructure required to make that revolution possible.
A new report from the United Nations University Institute for Water, Environment and Health shines a light on AI’s environmental footprint. As AI becomes woven into nearly every sector of society, from healthcare and education to business and government, it depends on a growing physical infrastructure that includes data centers, electricity grids, cooling systems, water resources, land, and critical minerals.
The report makes a simple but important point. AI is not merely software, but a physical system with measurable environmental costs.
Key Findings from the U.N. Report
- Energy Use: The environmental cost of AI begins with electricity. Global data centers already consume enormous amounts of power, and demand is projected to rise sharply as AI workloads expand. The report estimates that data center electricity consumption could exceed 945 TWh annually by 2030, enough to rank among the world’s largest electricity users.
- Water Use: AI requires significant amounts of water to cool the servers that train and operate advanced models. Training GPT-4 is estimated to have consumed roughly 600 million liters of water, while GPT-5 could require nearly one billion liters. As AI adoption accelerates, growing water demand may place additional pressure on regions already facing water scarcity.
- Carbon Footprint: The report found that AI’s environmental impact extends well beyond energy consumption. Training GPT-5 alone is estimated to generate approximately 42,000 tonnes of carbon emissions. Because many data centers continue to rely on electricity grids powered by fossil fuels, AI’s carbon footprint remains closely tied to broader energy system challenges.
- Land Use: The physical infrastructure supporting AI requires substantial land for data centers, power generation, and supporting facilities. By 2030, the land footprint associated with global data center electricity demand could exceed 14,500 square kilometers, roughly twice the size of the Jakarta metropolitan area.
- AI in Everyday Use: While training frontier models receives the most attention, the report estimates that everyday use accounts for 80% to 90% of AI’s total energy demand. Billions of daily prompts, searches, images, and videos collectively create a much larger environmental footprint than many users realize.
- Environmental Justice: One of the report’s most important findings is that AI’s environmental costs are not distributed evenly. The benefits of artificial intelligence often flow globally, while the burdens associated with electricity demand, water withdrawals, land use, mining activity, and electronic waste are frequently concentrated in specific communities.
The Data Center Boom Is Reshaping the Planet
The report projects that global data center electricity consumption could exceed 945 terawatt-hours annually by 2030, enough to rank among the world’s largest electricity consumers.
The environmental footprint extends well beyond energy. Data centers require enormous amounts of water for cooling, land for development, and minerals for advanced computing hardware. By 2030, the report estimates that AI infrastructure could generate up to 2.5 million metric tons of electronic waste annually.
These findings reinforce themes explored in our previous Voices.Earth article, The Dark Side of AI Data Centers: What Few People See Behind the Boom. As demand for artificial intelligence accelerates, so does the environmental footprint supporting it.
One Planet. One Domain.
The findings of this report reinforce a simple reality. There is no digital economy separate from the physical planet.
Every AI query, every data center, and every technological breakthrough ultimately depends on energy, water, land, and natural resources drawn from a single interconnected Earth system.
That reality sits at the heart of the One Planet. One Domain. message that continues to guide the .earth community.
Organizations using a .earth domain signal their commitment to sustainability, transparency, and long-term stewardship while joining a growing global network focused on building a more resilient future.
Registrar Partners That Participated in the Earth Day 2026 Promotion
- GoDaddy (Global)
- Gonbei Domain (Japan)
- Ledl.net (Europe)
- NETIM.com (Global)
- Openprovider (Global)
- OVH (France/Global)
- OwnRegistrar (Global)
- Porkbun (Global)
- COREHub (Europe)
- CentralNic Reseller (Global)
- Lexsynergy (Global)
- DomainCostClub.com (US) (Global)
- Realtime Register B.V.
These partners helped bring the Earth Day 2026 promotion to a global audience, making it easier for organizations and individuals to align their digital identity with a more sustainable future.
To learn more and get started, visit Voices.Earth, and check out the latest Voices.Earth podcast series for inspiring stories from our community.
