Global Data Centers Hit 415 TWh in 2024 as AI Workloads Drive Record Energy Demand
Global data centers consumed approximately 415 terawatt-hours of electricity in 2024 — about 1.5% of all power generated worldwide — with AI workloads accounting for roughly 15% of that total. The International Energy Agency's 2025 Energy and AI report projects consumption will more than double to 945 TWh by 2030 in its base-case scenario. In the United States, data centers consumed around 183 TWh in 2024, representing 4% of national electricity use, and accounted for roughly half of all new U.S. electricity demand growth in 2025. AI inference and training hardware is growing at 30% per year in electricity demand — more than three times faster than conventional servers.
claim: Global data centers consumed approximately 415 TWh of electricity in 2024 — roughly 1.5% of worldwide supply — with AI infrastructure accounting for 15% of that demand and driving projections of 945 TWh annually by 2030.
Sources · prominence score
Evidence Quality
Tier Mix
Pipeline Warnings
- Unknown source host — defaulted to T? (lowest credibility)CredibilityScorer · epri.com
- Unknown source host — defaulted to T? (lowest credibility)CredibilityScorer · pewresearch.org
- insufficient_candidatesAlgox:topK · 5/6
Findings
- Global data centers consumed 415 TWh in 2024 — 1.5% of world electricity — growing at 12% per year over five years, per the IEA.
- AI servers accounted for 15% of total data center energy demand in 2024 and are growing at 30% annually, versus 9% for conventional servers.
- The IEA base case projects global data center electricity will reach 945 TWh by 2030; the high scenario reaches 1,100 TWh — nearly triple 2024 levels.
- U.S. data centers consumed ~183 TWh in 2024 (4% of national electricity) and accounted for roughly 50% of all new U.S. electricity demand growth in 2025.
- EPRI and LBNL both project U.S. data center electricity could reach 6.7–12% of total national supply by 2028–2030, driven primarily by AI infrastructure.
Grid operators, utilities, and technology procurement teams need independent verification of AI's power footprint before committing to infrastructure investments. UVRN checks whether published projections from the IEA, national laboratories, and independent researchers actually converge — or whether the uncertainty range is too wide to act on.
- Run ID
- run-080
- Agent
- [email protected]