WASHINGTON — The rapid expansion of data centers to support artificial intelligence is driving a surge in electricity and water consumption, raising environmental concerns worldwide, according to recent reports.
Global data centers consumed about 448 terawatt-hours of electricity in 2025, producing around 208 million tons of carbon dioxide emissions, a United Nations University study released this week found. That electricity use rivals the consumption of many large countries, while associated water use reached about 1.2 trillion gallons.
By 2030, electricity demand from data centers is projected to nearly double to approximately 945 terawatt-hours, accounting for nearly 3 percent of global electricity use, the report said. AI is expected to drive about 40 percent of that total, up from around 20 percent currently.
The facilities require vast amounts of water for cooling servers. A single large AI-focused data center can consume up to 5 million gallons per day, equivalent to the usage of a small city, according to environmental analyses.
The UN University report estimated the associated water footprint could reach 9.3 trillion liters by 2030, comparable to the basic annual domestic needs of 1.3 billion people in Sub-Saharan Africa. Land use impacts from related power infrastructure were also highlighted.
Industry representatives noted efforts to mitigate impacts through renewable energy procurement and efficiency improvements. Tech companies including Microsoft, Amazon and Google have set carbon-free energy goals, though some reports indicate challenges in meeting them amid rapid growth.
"The demand is enormous," an expert involved in the UN study said, referring to the scale of AI-driven requirements.
Community opposition has grown in multiple locations, with concerns over local power grids, water resources and air quality from backup generators. In the United States, data centers already account for a notable share of electricity demand, with projections showing significant increases.
The International Energy Agency and other analysts have pointed to the dual nature of AI, which can also help optimize energy systems, though current infrastructure strains remain prominent.
Regulators and utilities in various regions are assessing grid capacity and resource allocation as construction continues. Details on specific mitigation measures and long-term projections continue to evolve with ongoing research and policy discussions.
The situation underscores the growing intersection of technological advancement and resource management as AI adoption accelerates.
