In April 2024, Princeton assistant professor Jesse Jenkins argued that AI-related electricity demand did not look like a U.S. grid crisis when compared with historical demand growth. That was a dated judgment, not a guarantee about today’s grid. Federal estimates since then show rapidly rising data-center use, while leaving unanswered how much is specifically due to AI and whether power and transmission will be ready in the regions where facilities are built.
What did the Princeton professor say?
At a Society of Environmental Journalists event at Princeton in April 2024, Jenkins told Bloomberg News, “It’s not a crisis.” He was comparing the potential increase in electricity demand from data centers with the pace of U.S. demand growth in other periods—not claiming that every grid could accommodate new facilities without difficulty.
Bloomberg reported that Jenkins expected data centers could raise annual U.S. electricity-demand growth to four times its recent average. The recent average he cited was 0.04% a year, compared with 2.3% a year from 1980 through 2005. His point was that a multiple of a very low recent rate could still be modest beside historical growth. Those figures describe the comparison he made in 2024; they should not be treated as a current forecast.
How much electricity do U.S. data centers use?
Federal estimates concern data centers as a whole, not AI workloads alone. The Department of Energy’s announcement of a 2024 Lawrence Berkeley National Laboratory report gives these national figures:
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
#1 Best Overall
| Measure | Estimate | What it describes |
|---|---|---|
| Electricity use in 2014 | 58 TWh | U.S. data centers |
| Electricity use in 2023 | 176 TWh, about 4.4% of total U.S. electricity | U.S. data centers; 2023 use estimate |
| Projected electricity use in 2028 | 325–580 TWh, about 6.7%–12% of total U.S. electricity | Range estimated in the 2024 report; not an observed result |
TWh means terawatt-hours, a measure of electricity used over time. The range for 2028 is broad; it is an estimate, not a single settled outcome. And because it covers all data centers, it cannot be read as an estimate of AI’s separate share.
What has changed since the 2024 assessment?
Overall U.S. electricity demand is growing faster
The Energy Information Administration reported in 2026 that total U.S. electricity demand grew about 1.7% per year from 2020 to 2025, compared with 0.1% per year from 2005 to 2019. These rates cover total electricity demand, not AI alone, so they cannot show how much of the acceleration data centers caused.
Newer data-center estimates point to continued growth
The Department of Energy and Lawrence Berkeley National Laboratory’s 2025 update estimated, in its reference case, that data-center electricity consumption grew 22% from 2024 to 2025 and would grow 29% from 2025 to 2026. These are estimates for data centers overall, not measured growth in AI power use alone.
Long-range figures are scenarios, not promises
In its 2026 Annual Energy Outlook scenarios, EIA put data-center server electricity consumption in 2050 between 446 billion and 818 billion kWh. The high end assumes faster growth in server power draw and the installed server stock. The range signals how much long-run results depend on assumptions; it is not a guaranteed outcome or a forecast for AI alone.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Will AI data centers strain the power grid?
National annual electricity totals cannot answer that by themselves. A country may have enough generation in aggregate while a particular region faces trouble serving a large new load at the time and place it arrives. The Department of Energy describes data-center demand as fast-growing, geographically uneven, and dependent on reliable power. Local effects turn on where facilities connect, when their demand comes online, and whether generation and transmission capacity are available in time.
Forecasts also change as AI uses, efficiency, and data-center plans evolve. The reviewed federal figures do not establish a precise future share of data-center consumption attributable specifically to AI, separate from other workloads and drivers of electricity demand. It would therefore be inaccurate to assign all projected data-center growth to AI.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What does this mean for household electricity bills?
These national estimates do not establish whether a household’s bill will rise because of AI data centers. That outcome depends on local utility and grid conditions, costs, and how new supply and infrastructure are provided—questions the cited national figures do not settle for every region. A large national growth estimate alone is not enough to predict a customer’s rate or bill.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errors
