Global electricity demand driven by artificial intelligence (AI) is projected to rise sharply in the coming years before stabilizing as technological advancements improve energy efficiency, according to a report released at the 2025 Global Energy Interconnection Conference in Beijing. The study, published by the Global Energy Interconnection Development and Cooperation Organization (GEIDCO), highlights AI's dual role as both a driver of energy consumption and a catalyst for innovation in sustainable tech.
The Surge in AI-Driven Energy Use
Data centers worldwide consumed approximately 415 billion kilowatt-hours of electricity in 2024 – four times higher than 2005 levels – with AI applications accounting for 1.5% of global electricity use. Accelerated computing servers, specialized for AI tasks, saw their energy consumption grow by 36% annually between 2020 and 2024.
Factors Influencing AI’s Energy Footprint
The report identifies three key variables: software algorithms, computing hardware, and data center efficiency. Advanced energy-management systems in data centers already reduce cooling-related energy use by 15%, while intelligent computing is expected to dominate future power demand as AI permeates industries from healthcare to finance.
Balancing Growth and Efficiency
While AI will drive localized spikes in electricity demand, the report predicts a logarithmic growth pattern: rapid initial expansion followed by stabilization. This trend reflects improvements in computing technologies and energy optimization strategies, suggesting AI's global energy impact will remain manageable despite regional challenges.
