Cornell Study Finds Data Centers Pay 38% of Grid Upgrade Costs, Leaving Rural and Low-Income Residents With the Bill

Data centers paid only 38% of the grid infrastructure costs associated with their growth in 2025, leaving residential ratepayers to cover the remaining 62%, according to a Cornell study published Nov. 9, 2026, in Nature Energy. The finding matters because it reframes the AI data-center boom as a cost-allocation problem, not just an electricity-demand problem: the people who benefit least from new data centers can pay the most for the transmission upgrades those facilities require.

What the Cornell study measured

The Cornell researchers analyzed energy-demand growth and infrastructure projects across four major U.S. power grids from 2020 to 2025. The grids cover regions with the largest data-center clusters: the southern San Francisco Bay area, north Texas around Dallas-Fort Worth, the Midwest, and the mid-Atlantic region including northern Virginia. The team used data from 4,000 existing and proposed data centers, power-price records, utility rate filings, power-line plans, and federal household and job data.

The study reports that data centers paid $3.12 billion of an $8.25 billion infrastructure bill in 2025—38% of the total. The remaining $5.13 billion, or 62%, fell largely on residents. For the lowest-income households, extra energy costs reached up to 28% of income.

Transmission investment in energy infrastructure projects rose from $2 billion in 2020 to $34 billion in 2025, with data centers driving much of the increase, according to the study. Data centers are projected to consume between 8% and 17% of U.S. electricity by 2030, with 70% to 80% of that demand attributed to AI.

Why residents absorb the costs

The core problem is that infrastructure costs are often allocated using past energy consumption. Data centers create new demand, but before they are built they have no consumption history. That creates what Fengqi You, the Roxanne E. and Michael J. Zak Professor in Energy Systems Engineering at Cornell, called a chicken-and-egg problem.

“Because if you’re going to build a data center, you need to expand this power infrastructure. Otherwise, there’s no electricity for them to run. But before they build, they’re not using energy. So with this model of relying on past consumption, households end up taking on much of the infrastructure costs.”

The study is described as the first multi-region, long-term assessment of how data-center development affects infrastructure growth and how costs are distributed across communities. You said the research was driven by a simple question: “With the booming of data centers, who pays for that expansion and who benefits?”

The price effects are regional

Data-center growth raised wholesale electricity prices—the amounts utilities pay regional power markets before delivering electricity to homes—by nearly 10% or more in the most-affected hubs by 2025. Virginia saw a 15% increase, and California’s southern Bay Area saw a 20% increase. In New York and New England, where data-center construction has been limited, residents have not seen clear price increases due to data-center expansion.

The cost burden also crosses state lines. Residents in New Jersey, Pennsylvania, and Ohio help pay for grid upgrades linked to data-center growth in Virginia, the researchers found.

“You do not have to live next to a data center to help pay for its power infrastructure. The grid connects communities across states, and cost-sharing rules connect them financially.”

Rural and lower-income residents are disproportionately affected. The share of rural families shouldering a high energy burden—spending more than 6% of income on energy—rose 76% more than among urban residents. The authors also cite a national survey finding that rural residents are less likely to use AI tools such as ChatGPT than urban residents.

The benefits side: jobs and state incentives

The Cornell study does not treat data centers as purely a cost. Each gigawatt of data-center capacity can create 15,000 jobs across the economy, with an average of 2,700 jobs directly associated with the data center, though not necessarily all on-site.

State incentives for data centers produced returns on investment, if inconsistently. California reaped $180 in worker income for every dollar spent, while Virginia increased income by $2 for every dollar invested. The researchers found that the presence of an existing, qualified workforce was a stronger indicator of high returns than the amount of incentive spending.

“In a lot of discussions right now, people take one side or the other, completely yes on data centers or completely no. But there are benefits and costs, and the key issue in the end is how to balance the trade-offs.”

What the authors recommend

The study recommends changing cost-sharing rules and reducing peak demand. In all four regions, the researchers found that if data centers cut back power use during 22 of the region’s peak demand hours each year, they would relieve strain on the grid and minimize the need to expand. Tech companies that own multiple data centers could shift operations to another region during peak demand.

“It’s feasible, and the challenge is more about implementation. Coordination is needed, but there’s a clear benefit for better coordination.”

The authors also suggest that states charge data centers directly in line with the new demand they add, or create programs—like a recent Minnesota law—to direct funds to low-income households to offset higher energy expenses. Tax incentives could be tied to job creation and increased wages so that benefits are more assured.

The policy context is moving. Six states have recently set special terms for very large energy consumers, and in March, seven major technology companies signed the White House Ratepayer Protection Pledge, a commitment to pay for required grid updates.

The Hacker News debate: methodology, supply constraints, and who captures the benefits

Hacker News commenters largely accepted that data centers are changing electricity costs, but they disagreed sharply about the cause and the study’s methodology.

One prominent critique came from commenter Aurornis, who said the authors shared a replication package but used a misleading method. Aurornis argued that the study assigns transmission-upgrade costs in proportion to peak energy demand rather than to the actual cost of the infrastructure upgrade.

“They’re basically taking the cost of transmission infrastructure upgrades and then assigning the cost of those upgrades proportional to peak energy demand of the customer base, not the cost of the infrastructure upgrade.”

Aurornis added that connecting 100,000 new residential addresses can cost more than a couple of data centers close to a power station, and that large customers are billed for infrastructure upgrade costs after utilities run their analyses.

“This paper is ignoring the utilities’ analyses and substitutes their own highly simplified proportional-to-peak-demand estimates, and that’s how they get their headline number.”

Aurornis also flagged a finding that cut against the expected narrative: the study shows data-center construction incentives to be a good deal in some states.

Other commenters focused on supply and permitting. MostlyStable argued that increased demand raises prices under a fixed-supply regime, but that fixed supply is a policy choice rather than a physical requirement.

“I would argue that it is not the data centers driving up power prices, it is our policies around how difficult it is to provision new power (or build almost anything at all) in the US.”

MostlyStable pointed to a bipartisan bill, BAAJA, as a partial fix. Commenter tonymet made a related argument: the amount of power needed by data centers is small compared with the generation gap, and rates are rising because no baseload generation has been added for more than 15 years in many markets.

“Rates aren’t going up because of datacenters. They are going up because no investment has been made for more than 15 years, and we have been offsetting the ‘crunch’ with creative workarounds.”

Another thread asked why data centers are treated as a problem rather than an opportunity. Hathawsh suggested that cities and counties could charge high property taxes to data centers and return the benefits to residents.

“Cities and counties can charge a high property tax to data centers and the residents can benefit. They might even give residents a dividend from the data center property tax.”

Bshaughn, writing from Chicago, said the cost is broader than low-income and rural ratepayers: residents with a strong grid and relatively cheap power are also seeing bills rise, while utility workers face long hours and denied raises.

“Its insane how few people are actually benefiting from this, and how simply taxing these data centers appropriately would fix it.”

Austin-cheney framed the issue as a local-benefit gap, citing a survey that found most people would rather have a nuclear power plant in their backyard than a data center.

“Of course everybody would rather have nuclear power plant in their backyard because then they get cheap, clean, reliable electricity. They get nothing from the data center except competition for basic utilities.”

Arjie offered a blunter version of the same concern: “Datacenters are this generation’s nuclear power plants. Our kids will look back and wonder what we were thinking.”

Jonhohle described a regulated, for-profit utility that made $430 million in profit and was asking regulators to raise rates “only” 14%, while the price it pays for residential solar approaches zero. Daishi55 argued that the real problem is political influence: negative externalities land on people least able to buy influence. Pseudonimous-pr noted a coverage coincidence: crypto data centers grew rapidly from 2010 to 2023 with relatively little mainstream criticism, while AI data centers now receive intense scrutiny.

What the study changes—and what it does not

The Cornell study provides a concrete estimate for a debate that has often relied on anecdote: in 2025, data centers covered 38% of the $8.25 billion in infrastructure costs linked to their growth, while residents covered 62%. It also documents regional price effects, cross-state cost-sharing, and a heavier burden on rural and low-income households.

The study does not settle the methodological question. Aurornis’s critique—that actual utility cost-of-service analyses may tell a different story than a peak-demand allocation model—is a direct challenge to the headline number. Nor does the study resolve the supply-side argument: commenters who blame permitting and generation shortages for higher prices see data centers as a symptom rather than the root cause.

What the study does establish is that policy design matters. Cost-sharing rules, peak-demand incentives, direct charges for new demand, low-income offsets, and tax-incentive conditions all change who pays and who benefits. As You put it, the question is not only whether the U.S. can generate enough electricity, but whether “the rules for paying for the grid are keeping pace with who is driving new demand.”

Sources