AI Data Centers’ Carbon Footprint Rivals Small European Nations, Study Reveals

Artificial intelligence (AI) technology is on track to become a significant environmental concern. A recent study forecasts that by 2025, the carbon footprint of AI data centers could rival that of small nations or major cities. The emissions from these centers are estimated to range between 32.6 million and 79.7 million tonnes of carbon dioxide, comparable to New York City’s 52.2 million tonnes recorded in 2023.

AI Data Centers and Their Environmental Impact

The study revealed that the environmental effects of AI data centers extend beyond carbon emissions. The amount of water these facilities consume could mirror the global annual bottled water industry, with an estimated use between 312.5 and 764.6 billion liters in 2025. This total accounts for both direct cooling efforts and indirect water consumption related to electricity production.

Regional Emission Comparisons

• New York City: 52.2 million tonnes CO₂ in 2023.
• Norway: Approximately 31.5 million tonnes CO₂ in 2023.

Data centers, which support cloud computing and streaming services, are increasingly in demand, pushing energy and water usage to new levels.

Europe’s Cleaner Energy Advantage

Europe houses about 15 percent of the world’s data centers, while the United States accounts for roughly 45 percent. The carbon intensity of Europe’s power grid is around 174 grams of CO₂ per kilowatt-hour, significantly lower than the global average of 445 gCO₂/kWh and the United States’ 321 gCO₂/kWh. This indicates that data centers in Europe have a reduced carbon footprint compared to their counterparts in other regions.

Lack of Transparency Among Technology Companies

The study highlights a concerning lack of transparency in environmental reporting by major tech firms. None of the nine companies analyzed, including Amazon, Apple, Google, and Meta, provided AI-specific environmental metrics, despite acknowledging that AI is a growing factor in energy consumption.

Research methods combined public sustainability reports with estimates of AI electricity demands, yet the authors noted that many disclosures do not differentiate between AI and non-AI computing undertakings. This lack of clarity complicates the assessment of AI’s environmental impact.

Calls for Improved Environmental Reporting

To better manage the ecological footprint of AI systems, the study emphasizes the need for enhanced transparency and policy changes mandating companies to disclose specific environmental metrics. This includes operational scale, site locations, and water usage effectiveness for AI facilities.

As data centers’ environmental impact continues to rise, increasing transparency within the tech industry is crucial for mitigating adverse effects and fostering sustainable practices.