AI and data centers leave goliath-sized environmental footprints globally

Key Points:

• A United Nations University report reveals that global data centers consumed 448 trillion watt-hours of electricity last year, producing 208 million tons of CO2 and using 1.2 trillion gallons of water, with environmental impacts comparable to some of the world’s largest countries.
• Data center energy use is expected to double by 2030, reaching 935 trillion watt-hours and producing nearly 440 million tons of CO2, driven largely by artificial intelligence, which currently accounts for 20% of their energy use and is projected to rise to 40%.
• The report highlights the hidden environmental costs of AI, emphasizing that AI operations require significant physical infrastructure and energy, despite lacking visible pollution, and suggests that more concise AI queries could reduce energy consumption by 25%.
• Industry representatives acknowledge environmental concerns but stress ongoing efforts to improve efficiency and responsible growth, while experts warn that increased efficiency often leads to greater overall energy use due to higher demand.
• Transparency issues remain a challenge, as many companies do not disclose detailed data on their energy consumption or data center operations, complicating efforts to manage and mitigate environmental impacts.