RESEARCH
AI Workloads Set to Double Global Data Center Power Demand by 2030
The IEA's 2025 Energy and AI report documents that global data center electricity consumption reached 415 TWh in 2024 — roughly 1.5% of total global electricity use — and projects this nearly doubling to approximately 945 TWh by 2030, driven primarily by AI workloads. Multiple independent research institutions confirm the doubling trajectory, though projections vary by scope and methodology. The Delta Engine finds indeterminate consensus across five sources: coverage prominence varies significantly between institutions that published full dedicated reports on this finding and those that feature it as supporting analysis. AI data centers' share of total data center power is projected to rise from roughly 10% today to 35–50% by 2030.
Findings
- IEA: Global data center electricity use hit 415 TWh in 2024 (~1.5% of world electricity); IEA projects ~945 TWh by 2030 — a near-doubling in six years.
- LBNL/DOE: US data centers alone consumed 176 TWh in 2023 and are projected to reach 325–580 TWh by 2028, driven almost entirely by AI training and inference expansion.
- Goldman Sachs projects a 165% increase in global data center power demand from 2023 to 2030, requiring roughly $720 billion in grid investment and adding 215–220 million tonnes of CO₂ through 2030.
- AI workloads are projected to grow from ~5–15% of total data center power today to 35–50% by 2030, meaning AI alone will consume more electricity than all of today's data centers combined.
Delta Engine result
↔ Divergence Detected — Δ 0.1573 (threshold 0.05)
- IEA T5 96
- LBNL / DOE T5 94
- S&P Global T3 85
- Carbon Brief T3 82
- Goldman Sachs T0 91
Evidence quality
avg 71 · min 25 · max 95 · spread 70