Data Centers & Energy
Data centers consumed about 415 TWh of electricity in 2024, roughly 1.5% of global power demand (IEA). The IEA projects this will more than double to about 945 TWh by 2030, nearly 3% of world electricity, as AI training and inference loads scale. The binding constraint is increasingly grid interconnection and reliable power supply rather than capital or land.
Key insights
AI is bending the curve
Through 2018 data-center energy demand was roughly flat — efficiency gains offset traffic growth. The post-2022 shift to large transformer training and inference reverses that. The IEA expects electricity demand from AI-focused data centers to as much as quadruple by 2030.
Grid interconnection is the bottleneck
In northern Virginia (the world's largest data-center cluster) Dominion Energy quotes 4–7 year interconnection waits for new gigawatt-scale loads. Similar queues in Dublin, Frankfurt and Singapore have triggered de facto moratoriums on new builds. Microsoft, Amazon, Google and Oracle are signing long-term PPAs for solar+storage and increasingly for nuclear (small modular reactors, restarts of retired plants).
Water use is locally significant
Cooling consumes 1.7 to 9 litres of water per kWh of IT load depending on climate and design. A single hyperscale campus can use as much water as a town of 50,000. Direct-to-chip liquid cooling and air-side economisers reduce water use but increase electricity demand — there's no free lunch.
Global data-center electricity demand 2010–2030 (forecast)
TWh per year, IEA central case
Show data table
| TWh | |
|---|---|
| 2010 | 200 |
| 2015 | 210 |
| 2018 | 210 |
| 2020 | 260 |
| 2022 | 350 |
| 2024 | 415 |
| 2030 (IEA projection) | 945 |
Key Finding: Demand was roughly flat 2010–2018 despite massive traffic growth, then accelerated. The 2030 point is an IEA projection of about 945 TWh, slightly more than Japan uses today.
Average PUE — hyperscale vs enterprise (2015–2026)
Power Usage Effectiveness, lower is better
Show data table
| Hyperscale | Enterprise (avg) | |
|---|---|---|
| 2015 | 1.3 | 2.2 |
| 2018 | 1.25 | 1.85 |
| 2020 | 1.2 | 1.7 |
| 2022 | 1.18 | 1.66 |
| 2024 | 1.18 | 1.62 |
| 2026 | 1.2 | 1.58 |
Key Finding: Hyperscale operators sit near the practical floor of 1.10–1.20; enterprise data centers lag at about 1.6 on average.
Methodology & caveats
PUE explained
Power Usage Effectiveness = total facility power / IT equipment power. PUE 1.0 is the theoretical lower bound (no overhead). Hyperscalers report 1.10–1.20 at modern sites; the enterprise average sits near 1.6. PUE is a useful efficiency proxy but ignores water and carbon.
What 'AI energy use' includes
Training a frontier model can consume 50–100 GWh once. Inference (serving queries) is more distributed but in aggregate often exceeds training. Public figures generally cover training; total AI energy includes inference, data ingestion, fine-tuning and supporting infrastructure — and that aggregate is poorly measured.
Reporting gaps
Data-center energy is reported voluntarily through programmes like the EU Code of Conduct and operator ESG disclosures. National statistical agencies do not separate the sector cleanly from broader commercial demand. IEA estimates therefore rely on operator surveys plus modelled adjustments — confidence intervals are ±20% at the global level.
Why estimates vary so widely
The IEA's 415 TWh is a modelled estimate, because almost no country reports data-center electricity separately. Figures for the same year can differ by tens of TWh depending on whether crypto mining is included. Older IEA reports used a 460 TWh figure for 2022 that included crypto, which is why that number still circulates. Forecasts for 2030 span an even wider range, and the IEA itself publishes lower and higher scenarios around its base case of about 945 TWh.
Global share hides the local problem
At under 2% of world electricity, data centers look small. The strain is local: capacity clusters in a few grid zones, so a single campus can match a mid-size city's demand and take years to connect. What would change the outlook is chip efficiency, the pace of AI revenue, and how quickly on-site generation and new transmission can be built. The demand curve is an assumption about all three, not a trend that must continue.
Cite this page
GlobeChart. “Data Centers & Energy.” Last updated 2026-10-31. https://globechart.com/technology/data-centers-energy/
Please also credit the original publisher named under each chart: IEA Energy and AI (April 2025); Uptime Institute Global Data Center Survey, hyperscaler ESG reports. Spotted an error? Tell us.