The rapid proliferation of artificial intelligence is fundamentally reshaping global energy consumption. Data centers, which consumed approximately 415 TWh in 2024, are projected by the IEA to nearly double their electricity usage to roughly 945–950 TWh by 2030 [2] [3] [4] [15].
Escalating Power Demands
AI workloads are significantly more power-intensive than traditional computing; modern AI racks can require 50–100 kW compared to 5–10 kW for conventional hardware [20]. A 2026 study suggests that facilities run by six leading technology firms alone could account for 239–295 TWh by 2030—roughly 1% of total global electricity demand [5] [7].
Grid and Environmental Challenges
This growth is straining power grids, particularly in North America and Europe, where interconnection delays for new transmission infrastructure often exceed five years [13] [16] [20]. To bypass these bottlenecks, some operators are turning to behind-the-meter gas generation or nuclear power, raising concerns about local pollution and climate targets [9] [10] [16]. Beyond electricity, AI infrastructure creates significant environmental pressure, with projections estimating billions of liters of water consumption for cooling and millions of tonnes of e-waste annually by 2030 [11] [15] [18].
Sources
- Artificial intelligence data centers could reach one percent of …
- AI Data Centers Could Consume 1% of Global Electricity by 2030
- Environmental impact of AI – Wikipedia
- US Utilities Plan $1.4T for AI Data Centers: 27% Capex Surge [2026]
- The environmental impact of AI: Who bears the costs? DevelopmentAid
- Data Center Power & Energy News 2026: Weekly Project Updates
- Tarumahiman
- Green AI 2026: The Data Center Energy Crisis Explained
- AI datacentres pose growing threat to electricity systems worldwide …
- AI Infrastructure Growth: Power Demand Outlook