Assessing the Carbon Emissions and Energy Consumption of U.S. Hyperscale Data Centers

A comprehensive study of 403 U.S. hyperscale data centers reveals they consumed 68-99 TWh of electricity between May 2024 and April 2025, generating 37-54 million metric tons of CO2 emissions. The findings show HDC carbon intensity is 48% higher than the national grid average, driven by rapid AI infrastructure expansion and heavy reliance on fossil fuels.

The exponential growth of hyperscale data centers supporting artificial intelligence has created an environmental challenge that demands systematic assessment. This research provides the first detailed facility-level inventory of U.S. HDCs, establishing a baseline for understanding their true environmental footprint. The 68-99 TWh electricity consumption represents 1.8% of total U.S. electricity demand—a substantial and rapidly growing share driven by AI model training and inference workloads.

The carbon intensity disparity is particularly significant. At 545 gCO2/kWh, HDC operations generate emissions 48% above the national grid average, primarily because data center locations are not optimized for renewable energy availability. Many facilities operate in regions with higher fossil fuel penetration, and the sheer scale of demand outpaces local renewable capacity. With 54% of attributed generation sourced from fossil fuels, the data center industry faces pressure to accelerate renewable energy procurement and grid decarbonization.

This environmental assessment carries implications for both investors and policymakers. Tech companies building AI infrastructure face mounting scrutiny over sustainability claims, while regulators may increasingly link data center expansion permits to renewable energy commitments. The research provides a quantifiable basis for corporate environmental claims and regulatory frameworks. Energy-intensive AI applications could face higher operational costs as carbon pricing mechanisms expand.

The critical question ahead involves whether data centers can rapidly transition to renewable energy sources or whether grid decarbonization must precede further HDC expansion. Technology companies' net-zero commitments will be tested by actual emissions data, and investors should monitor whether facility-level improvements in energy efficiency and renewable sourcing emerge.

• →U.S. hyperscale data centers consumed 68-99 TWh of electricity in 2024-2025, representing 1.8% of total U.S. electricity demand.
• →HDC carbon intensity averages 545 gCO2/kWh, significantly exceeding the U.S. grid average of 370 gCO2/kWh by 48%.
• →Approximately 54% of attributed electricity generation for HDCs comes from fossil fuel sources, creating environmental compliance concerns.
• →The study analyzes 403 facility-level data centers, providing the most comprehensive U.S. inventory and attributional assessment to date.
• →AI-driven hyperscale expansion creates pressure for renewable energy procurement acceleration and grid decarbonization initiatives.