AI resource consumption triggers environmental concerns

The environmental costs of artificial intelligence, specifically regarding electricity and water consumption, have come under scrutiny following viral social media trends. Research from the United Nations University Institute for Water, Environment and Health indicates that generating a single AI image consumes enough energy to power a 10-watt LED bulb for approximately 17 minutes, with a water footprint of about 29 milliliters.

When scaled to a global level, the impact is substantial. Producing 1 billion AI-generated images could require approximately 29 million liters of water. The environmental footprint extends beyond individual image generation to include the operation of massive data centers, which require significant electricity for processing and vast amounts of water for cooling high-performance hardware.

The International Energy Agency (IEA) projects that global data center electricity consumption will rise from approximately 448 terawatt-hours in 2025 to about 950 terawatt-hours by 2030, driven largely by the expansion of AI. By 2030, the water footprint associated with the electricity used by AI-supporting data centers could reach 9.3 trillion liters, with a land footprint exceeding 14,500 square kilometers.