Do you think AI is free of cost? Artificial intelligence may be making work and life easier, but it’s also quietly driving up power bills across the U.S. The massive energy needs of AI data centres, the giant servers that run tools like ChatGPT, are straining electricity grids and leaving ordinary households to pay more.
Power demand from AI
A Bloomberg analysis shows that AI data centres now consume about 4% of U.S. electricity, a share expected to rise sharply as demand for generative AI accelerates. Despite more efficient chips and software, overall energy use keeps climbing because of the scale at which AI is expanding.
Regions with heavy data centre clusters are seeing the biggest jumps. Northern Virginia, parts of New York, and California are reporting power costs well above the national average. Liam Denning of Bloomberg explains, “Average annual power bills in all but four states within the PJM grid, plus DC, rose faster than the national average — and that was already over 6%.”
In New Jersey, bills climbed 13.3% in a year, while New Yorkers faced a 14.4% rise, both among the steepest increases in the country.
Why consumers are paying
Utilities say they need expensive upgrades to keep pace with AI’s power demand. The problem is, much of that cost gets passed down to regular families and small businesses. Monitoring Analytics, an independent grid watchdog, found that nearly two-thirds of new charges hitting bills this summer are linked to current and future data centre demand.
This has fuelled criticism that households are effectively subsidising Silicon Valley’s AI race. As Denning puts it, “It would be absurd to abandon the benefits of decades of deregulation just to accommodate data centers. At a minimum, they should bring their own power by committing to new plants.”
A flawed power billing system?
Energy experts also say the current pricing system is flawed. Utilities analyst at Sector and Sovereign Research, Hugh Wynne, compared it to a “rain dance,” pointing out that billions in extra charges have been collected but have delivered very little new power supply. That means homeowners are paying more without seeing a stronger grid in return.
What comes next?
Lawmakers and regulators are now pushing back. Proposals include charging data centres higher rates, forcing them to buy more renewable energy, or even letting states step out of the current pricing system. But such fixes will take time.
For now, the reality is clear. Electricity prices are rising faster than inflation, and AI’s growing appetite for energy is a major reason. Until new solutions are in place, households may continue footing the bill for the AI boom.
