Why electricity bills have become a central fight over data centers

A data center can pay every power invoice it receives. Households may still be paying part of its wider bill. The political question is how to divide the cost of the system serving it. That includes power plants, transmission lines, substations and reserves.

Ohio offers a concrete example. In July 2025, the Public Utilities Commission ordered AEP Ohio to create a data-center-specific tariff. The commission said its adopted settlement protected other customers from the risk of paying for investments that would be underused. It also directed the utility to lift its connection moratorium. That combination matters: the regulator treated consumer protection and continued construction as potentially compatible. PUCO announcement.

The dispute deserves attention beyond a simple argument for or against artificial intelligence. A community can support digital infrastructure while rejecting a specific allocation of financial risk. A developer can reasonably request clear access to power while disagreeing with the contract a utility offers. Those positions are different from a demand to prohibit every new facility.

The difference between using power and reserving capacity

Electricity is both a product and a service. It must be available when needed. A customer’s annual consumption does not fully describe the infrastructure needed to serve its largest demand. A facility that expects to expand also asks planners to reserve space for future use.

This distinction explains why paying for electricity actually consumed may not resolve the political objection. If equipment is constructed for an expected customer and that customer never arrives, someone must absorb the remaining cost. The possibilities include shareholders, the departing customer, other customers or some combination determined by the contracts and regulatory decisions.

The reverse problem also deserves scrutiny. A utility may overstate demand and plan more infrastructure than it needs. Protection against a customer’s departure does not prove that the plan is sound. Consumer protection requires examining the forecast as well as the bill.

For readers following the broader backlash, 567politics’ election coverage describes the political setting. Election results alone cannot show the size of a utility subsidy or the financial effects of a single project.

What Ohio’s approach illustrates

AEP Ohio’s current application page describes a mandatory process for new data centers and expansions seeking service. It requires large applicants to provide specific site and load information and pay for a load study. Those administrative steps help distinguish a development proposal from an immediately deliverable electricity requirement. They do not guarantee that a proposed project will operate. AEP Ohio tariff and application information.

The key principle is that an expression of interest should not be treated as equivalent to an enforceable commitment. A company can investigate locations before making a final investment decision. The public interest problem arises when several tentative requests become the justification for infrastructure whose costs will outlast those requests.

An effective review therefore asks what evidence supports a forecast. Site control, financing, customer commitments and a realistic construction schedule are more informative than a headline investment figure. Confidential details may need protection. But confidentiality should not prevent regulators from explaining why they consider the aggregate forecast credible.

That question connects the electricity dispute to the series’ analysis of temporary moratoriums. A pause may create time to set out financial rules. But the usefulness of the pause depends on what work is completed during it.

The strongest argument for stronger safeguards

The consumer-protection argument is straightforward. Households cannot easily relocate to another electricity system when a large industrial project changes a utility’s investment program. They also cannot negotiate individually over transmission planning. Regulators therefore need to evaluate obligations on their behalf before the infrastructure becomes a sunk cost.

This argument does not require assuming that data centers are irresponsible. Even a well-managed business can change its plans as technology, financing or customer demand changes. Contracts exist partly because good intentions cannot eliminate commercial uncertainty.

An exit payment can address some of that uncertainty. Financial security can make a payment promise more credible. A minimum billing obligation can provide revenue when actual demand falls below an agreed level. None of those mechanisms is sufficient without knowing what costs it covers and whether the obligation survives the business failure.

The public needs to know who pays if plans change. That includes a delayed opening, a smaller project, bankruptcy or closure. The review should also explain what happens when a shared improvement serves both the data center and other customers.

The strongest objection to special treatment

A developer’s objection can also be large. A tariff that distinguishes one industry needs an explanation rooted in cost and risk. If two customers create comparable demands, a regulator should be able to explain why their obligations differ.

Overly rigid commitments could discourage a project that would have contributed revenue to shared infrastructure. They could also reward older facilities with more favorable terms while imposing heavier obligations on new competitors. Whether that happens depends on the tariff and market, rather than the label attached to the customer.

These concerns do not show that special tariffs are inherently unfair. They show why the review should connect each requirement to a measurable risk. A charge justified by stranded investment needs a stranded-investment analysis. A reliability requirement needs a reliability explanation.

The Ohio commission’s announcement records competing settlement proposals, including one involving technology companies and industry representatives. That history is a reminder that “business” and “consumers” are not each single negotiating blocs. Manufacturing customers may have different priorities from technology customers, while low-income households may face different effects from large commercial users. PUCO case background.

What “pay your own way” leaves unanswered

The phrase sounds conclusive but does not identify a cost boundary. Does it include the connection to the site, upstream transmission, new generating capacity or backup reserves? Does the calculation consider benefits the project provides to other customers? Over what period is the comparison made?

These questions also determine whether an apparent subsidy is real. A new customer could increase total spending while reducing another customer’s share of a fixed cost. Alternatively, a project could bring revenue but require still larger incremental expenditures. Neither possibility can be settled from the project’s capital budget alone.

The suitable comparison is between credible system scenarios with and without the development, using consistent assumptions. Forecast uncertainty should be visible. An estimate may look precise even when its opening date is a guess. A range that reflects this doubt can be a better guide.

The same discipline applies to tax-incentive claims. A large private investment is not a direct measurement of public benefit. Each argument requires a clear account of who pays, who receives revenue and what would have happened otherwise.

Local permission cannot settle every power question

A zoning board may decide whether a building fits a location. That decision does not show that a utility’s regional investment plan is affordable. A utility may be willing to connect a customer. That does not settle neighbors’ concerns about noise, land use or water.

This division of authority can frustrate communities because the decisions arrive through different proceedings. A permit hearing may discuss landscaping while the largest financial obligation sits elsewhere. A better public record links the proceedings and states which agency has authority over each disputed issue.

A local benefit agreement should also be assessed separately from electricity protections. Funding a community facility may be valuable. But it does not show that nonparticipating ratepayers are protected. Residents outside the host municipality may help pay for a regional grid investment without receiving the host municipality’s tax revenue.

That geographic mismatch is one reason the resistance can extend beyond immediate neighbors. It also helps explain why a project that satisfies a town council may still face objections at a state utility commission.

A practical test of the competing claims

Start with the approved tariff and the order behind it. Then check the public service terms, the utility forecast and the analysis of who pays. A company statement and a campaign speech can explain a position. But neither substitutes for those records.

The review should distinguish an adopted rule from a proposal and a contracted load from a developer inquiry. It should identify the service territory. Ohio’s approach is evidence about Ohio’s regulatory choices. It is not proof that every American utility follows the same arrangement.

There is no project-specific calculation in this article showing how much a household’s bill changed because of one data center. Assigning a dollar figure without that calculation would create a false impression of precision. The policy dispute is significant even where the final household effect remains uncertain.

The central test is whether the customer’s promises cover the costs and risks its project creates. Proven benefits also deserve credit. That test can support approval, stronger conditions or rejection depending on the evidence. It gives both supporters and opponents a proposition that can be checked.

Reporting scope

This is an explanatory analysis of publicly available regulatory information reviewed November 5, 2026. It does not measure national opinion or infer it from public-hearing attendance. Related reporting examines local zoning and neighborhood impacts and water disclosure.

Photo credit

Representative file photo: Victor Grigas / Wikimedia Foundation, via Wikimedia Commons, CC BY-SA 3.0. The source photo is unaltered; the site may crop its display.

Differing perspectives

How should data centers pay for the electricity infrastructure they require?

Each position is summarized as its supporters make it, with equal weight and in alphabetical order. A summary is not an endorsement; the evidence for each is weighed in the story above.

• Pause growth until safeguards are adequate

  Who holds it: Food & Water Watch

  New development should pause until a regulatory framework addresses environmental and social harms. The group argues that continued construction risks committing communities before adequate protections exist. Public rules should be clear before construction starts.

  In their words: Food & Water Watch ↗

• Require firm financial commitments

  Who holds it: PUCO’s adopted consumer-protection approach

  New connections should proceed under rules protecting other customers against the cost of underused infrastructure. The regulator presents development and consumer safeguards as compatible objectives. Public rules should be clear before construction starts.

  In their words: Public Utilities Commission of Ohio ↗