Electricity bills are rising across the United States, with average household utility costs increasing by $122 per month compared to 2020. This increase is putting additional strain on family budgets.
One factor behind these rising costs is the growing demand from artificial intelligence (AI) data centers. These facilities require significant amounts of electricity and are often located in areas such as Virginia’s “Data Center Alley.” This region shares the PJM power grid, which supplies electricity to 13 states including New Jersey, Pennsylvania, and Ohio.
The rapid growth of AI data centers has contributed to a surge in grid demand. In a recent PJM market auction, the price for capacity—fees paid to guarantee power during peak periods—increased by 833%. According to an independent monitor’s report, data centers were responsible for about three-quarters of this increase.
This situation means that households are indirectly subsidizing the expansion of AI infrastructure. Utility companies fund new infrastructure projects by raising rates for all customers. As a result, everyday consumers pay higher prices even if they do not use AI services.
However, academic research indicates that states experiencing the largest increases in electricity demand have sometimes seen lower overall prices. For example, North Dakota had a nearly 40% rise in demand but inflation-adjusted prices fell during that period. The main cost drivers for rising bills are not just energy generation but also transmission and distribution—the delivery of electricity through poles and wires—as well as investments needed to harden grids against extreme weather events.
The contradiction arises because steady and predictable load growth can help lower rates by spreading fixed costs among more users. However, when demand grows rapidly and unpredictably—such as with the recent boom in AI data centers—it can lead to congestion on the grid and force operators to buy more expensive power at short notice.
Kiran Garimella, Assistant Professor at Rutgers School of Communication and Information, writes: “The problem isn’t just the AI boom. It’s trying to run a 21st-century technology race on a 20th-century power grid, all while using an outdated regulatory playbook.”
To address these challenges, researchers funded by the New Jersey State Policy Lab plan to conduct a pilot study in Virginia and New Jersey. They will collect detailed zip-code level data from households near new data center developments using compensated data donations. The aim is to better understand how increased electricity demand affects individual families’ finances.
There are technical solutions available that could help relieve pressure on existing grids without waiting years for new transmission lines. These include implementing dynamic line ratings based on real-time conditions and optimizing how electricity is rerouted around bottlenecks—approaches shown in studies to be effective within one or two years.
Garimella concludes: “They are not a permanent replacement for new transmission, but they are the critical bridge necessary to build the next generation grid intelligently.”
