If you have opened an electric bill recently, you have probably noticed the same thing many households have: it is higher than it used to be. Nationally, average residential electricity prices rose by roughly 20 percent between 2021 and 2024, according to the U.S. Energy Information Administration.[1] In some places, the increase has been sharper.
Data centers have become an easy explanation. The story is intuitive. Artificial intelligence tools, cloud computing, streaming, online shopping, and nearly everything else we now do online depend on large buildings full of servers. Those buildings use a lot of electricity. A 2024 Lawrence Berkeley National Laboratory report estimated that U.S. data centers consumed 176 terawatt-hours of electricity in 2023, or about 4.4 percent of total U.S. electricity use, and projected that the share could rise to between 6.7 and 12 percent by 2028.[2]
It is reasonable for people to ask whether data centers are showing up in their monthly bills. That question is especially relevant for New Jersey. The state has a meaningful data center presence, sits in the PJM regional power market, and is tied into the same broader grid region as Northern Virginia, the largest data center cluster in the country.
My collaborators and I spent the last semester trying to answer a narrow version of that question with data: when a data center opens nearby, do residential electricity bills in that area go up?
The short answer is: not in a robust, clearly detectable way. At least not yet.
What We Studied
We built a state-ZIP-level panel covering 22,834 units across 24 states from 2014 through 2024. We linked those places to 2,277 operational data centers with valid ZIP codes and operational dates through 2024. The electricity bill data come from the Census Bureau’s American Community Survey, which asks households what they pay for electricity. The data center records come from public interconnection filings, which utilities and grid operators use when large new customers request service.
The basic comparison is straightforward: when a data center begins operating in a ZIP code, do household electricity bills in that ZIP code rise relative to places that did not receive a data center?
That comparison is not perfect. Data centers do not locate randomly. They tend to choose places with available land, grid capacity, favorable tax treatment, and relatively low electricity costs. We used several standard tools to make the comparison more credible: fixed-effects models, event studies, matched comparisons, dose-response tests, and a separate analysis at the utility service-territory level.
Studying this at the utility service-territory level is important because residential electricity rates are not usually set ZIP code by ZIP code. They are set across utility service territories. If a data center raises costs for a utility, those costs should generally be spread across many customers, not charged only to the ZIP code where the building sits.
What We Found
The first pass suggests that ZIP codes with a data center had monthly residential bills about $2.09 higher than ZIP codes without one. That is small relative to the average bill in the sample, which is about $164 per month. It is also a borderline result statistically, meaning it is not the kind of estimate I would want to build a strong policy claim around by itself.
The result got weaker when we asked a tougher question: what happens if we compare data center ZIP codes only to places that had similar electricity bills before the data center opened? That is a more apples-to-apples comparison. Under that test, the estimated increase fell to about $0.90 per month, and the result was not statistically significant.
We also checked whether the original estimate was partly picking up changes in local household income. That matters because higher-income households often use more electricity, and data centers may be located in places whose income profiles are changing. After accounting for income, the estimated increase was about $1.84 per month, again not statistically significant.
One pattern is worth taking seriously. Bills do not necessarily change the year a data center opens. Utility costs often move into residential rates slowly, through rate cases and long-term infrastructure planning. When we looked several years after a data center opened, we saw some positive estimates four to five years later. But that result is less certain because fewer places in the data can be observed for that long after a data center opening. And when we shifted to the utility territory level, where rates are actually set, the delayed pattern did not appear clearly.
That utility-level analysis is the most important check. When we aggregated bills and data centers to the level at which rates are actually set, we found no statistically significant effect. The unweighted utility-level estimate was about $0.35 per month. The household-weighted estimate was negative, about -$2.00 per month. Neither of these were statistically distinguishable from zero.
We also ran a spillover test: if a data center affects bills through the utility rate base, then other ZIP codes served by the same utility should also be affected, even if they do not host a data center themselves. That test did not support a simple rate-spreading story. The same-utility spillover estimate was negative, while the own-ZIP estimate was positive but not statistically significant.
Taken together, the results do not support the strong version of the claim that data centers have already caused large, localized increases in household electricity bills. The evidence is better described as weak, mixed, and small in magnitude.
Why These Results Makes Sense
The finding may sound surprising because data centers really do use a lot of power. But electricity bills are shaped by institutions, not just physics.
In most states, utilities recover costs through formal rate cases. A utility goes to the state public utility commission, documents its costs, and asks for permission to charge rates that cover those costs plus an allowed return. Residential rates are then set across a utility’s service territory.
That structure has two consequences.
First, even if a data center raises a utility’s costs, those costs are usually spread across many customers. A large load in one town does not automatically mean that only the households in that town pay more.
Second, data centers are not usually served like ordinary households or small businesses. They are large-load customers, often on separate tariff schedules with demand charges, high-voltage delivery, and other provisions that are supposed to reflect the cost of serving them. Public utility commissions do not always get cost allocation right, but they are explicitly responsible for deciding who pays for what.
Virginia illustrates both sides of the issue. A 2024 report by the Virginia Joint Legislative Audit and Review Commission found that data centers bring local tax revenue and economic activity but also create major electricity-planning challenges. The report projected that data center-driven costs could put upward pressure on future residential bills if rate design does not change.[3] That is a serious warning. It is also a forecast about what may happen as much larger loads come online, not evidence that every nearby household has already seen a large data-center-driven increase.
The Big Caveat: “Not Yet”
Our data ends in 2024 and that could be part of why we are not seeing significant effects.
Most of the data centers we have in our 2,000+ data center sample were the ‘traditional’ data centers set up before the current AI wave. The next wave of data centers is different. The facilities being planned for AI workloads are much larger than many of the data centers in our sample. Some proposed campuses are measured in hundreds of megawatts, and in some cases approach a gigawatt. Those are not ordinary commercial loads. They can require new substations, transmission upgrades, generation contracts, and changes to long-term utility planning.
Those costs also take time to reach customers. A utility does not build transmission overnight, and it does not recover the cost of a major capital project in a single billing cycle. Generation and transmission projects can take years to plan, permit, build, and place into rates.
That is why our finding should not be read as “data centers will never affect residential bills.” The better reading is narrower: from 2014 through 2024, during the pre-hyperscaler period covered by our data, we do not find robust evidence that data center openings produced meaningful increases in residential electricity bills. The future could look very different.
What This Means for New Jersey
For New Jersey, the main risk is probably not a single data center raising bills in the surrounding ZIP code. The bigger issue is regional.
New Jersey is part of PJM, the wholesale electricity market that serves all or parts of 13 states and the District of Columbia. PJM’s 2025/2026 capacity auction cleared at $269.92 per megawatt-day for much of the region, up from $28.92 in the prior auction.[4] That is a very large increase. PJM attributed the jump to a combination of tighter supply, higher demand, market rule changes, and declining offered capacity.[5] Independent analysts and consumer advocates have pointed to data center-driven load growth as one important contributor, especially in parts of the PJM footprint.[6]
That kind of regional capacity-market effect is not something a ZIP-level study is designed to isolate. But it can still matter for New Jersey households, because wholesale market costs eventually flow into retail bills.
The policy question for New Jersey is less “Did the data center down the road raise my bill this year?” and more “How do we make sure large new loads pay the costs they impose on the grid before those costs are socialized across everyone else?”
That is where state policy can matter. Regulators can require clearer large-load tariffs, stronger upfront payment requirements, better interconnection deposits, minimum demand commitments, and cost-allocation rules that protect residential customers if a project is delayed, downsized, or canceled. The goal should not be to stop all data center development but to make sure the costs and benefits are assigned transparently.
We Are Talking to Residents Next
The quantitative results in our analysis answer one question, but not the whole question.
Electric bills are personal. When a household sees its bill jump by $30 or $40, and then reads about a new data center nearby, the connection feels obvious. Sometimes that connection may be wrong. Sometimes it may be partly right but operating through a much larger regional market. Either way, perception matters. It shapes whether people trust utilities, whether they support new infrastructure, and what they expect regulators and elected officials to do.
That is why the next stage of this work will focus on New Jersey residents. We want to sit down with households, look at what they are seeing in their bills, and understand how they explain those changes. The point is not to tell people that their concerns are misplaced. The point is to learn what people are experiencing, what information they trust, and where the gap is between system-level data and lived experience.
That gap matters for policy. If residents believe they are being asked to subsidize the AI buildout, regulators need to be able to show clearly whether they are or are not. If data centers are paying their fair share, that should be visible. If they are not, the rules should change before the largest projects are fully online.
For now, the evidence suggests that data centers have not yet produced large, detectable increases in local residential electricity bills. But “not yet” is not the same as “nothing to worry about.” New Jersey still has time to design rules that protect ratepayers before the next wave of electricity demand arrives at full scale.
References:
[1] U.S. Energy Information Administration, Electric Power Monthly, Table 5.6.A, “Average Price of Electricity to Ultimate Customers by End-Use Sector.” https://www.eia.gov/electricity/monthly/
[2] Arman Shehabi et al., 2024 United States Data Center Energy Usage Report, Lawrence Berkeley National Laboratory, LBNL-2001637, December 2024. https://energyanalysis.lbl.gov/publications/2024-lbnl-data-center-energy-usage-report
[3] Virginia Joint Legislative Audit and Review Commission, Data Centers in Virginia, 2024. https://jlarc.virginia.gov/landing-2024-data-centers-in-virginia.asp
[4] S&P Global Commodity Insights, “PJM power capacity auction clears at record high price of $269.92/MW-day for most of footprint,” July 30, 2024. https://www.spglobal.com/energy/en/news-research/latest-news/electric-power/073024-pjm-power-capacity-auction-clears-at-record-high-price-of-26992mw-day-for-most-of-footprint
[5] PJM Interconnection, 2025/2026 Base Residual Auction Report, July 30, 2024. https://www.pjm.com/-/media/DotCom/markets-ops/rpm/rpm-auction-info/2025-2026/2025-2026-base-residual-auction-report.ashx
[6] Synapse Energy Economics, Drivers of PJM’s Capacity Market Price Surge and the DC Impact, prepared for the Office of the People’s Counsel for the District of Columbia, May 2025. https://opc-dc.gov/wp-content/uploads/2025/05/PJM-Capacity-Market-Report-FINAL-OPC-Synapse.pdf
[7] Mike Rogoway, “Oregon’s data centers want a lot more electricity. Who’s going to pay? It could be you,” The Oregonian/OregonLive, October 2024. https://www.oregonlive.com/silicon-forest/2024/10/oregons-data-centers-want-a-lot-more-electricity-whos-going-to-pay-it-could-be-you.html
[8] Robert Baker, “Data centers will drive up electric rates for Georgians,” The Atlanta Journal-Constitution, December 6, 2024. https://www.ajc.com/opinion/opinion-data-centers-will-drive-up-electric-rates-for-georgians/QVFJCW76MVBSBBEWIGTGYHOZUM/
[9] Kiran Garimella, Do Data Centers Raise Residential Electricity Prices? Evidence from 24 U.S. States, working paper, March 2026. https://gvrkiran.github.io/content/AI_data_centers_electricity_bills.pdf