Cost Shift

Every new plug sends a bill somewhere. AI data centers can grow without making ordinary families pay higher electric bills only if the meter proves it.

The arithmetic is stubborn. Data centers use real power. New power lines cost real money. The companies that need the new power say they will pay for the power, the wires, and the risk.

The White House Ratepayer Protection Pledge names the terms. Companies should build, bring, or buy the energy their new data centers need. They should pay for required delivery upgrades. They should negotiate separate rate structures. They should pay for that power and infrastructure even if they do not use all of it.

No slogan there. A rule for assigning a bill.

The rule runs this way: new load creates new cost. New cost needs a named payer. If the payer is not the company that needs the load, the cost has shifted toward everyone else.

The Pledge Is About Assignment

An AI data center draws far more power than a laptop on a kitchen counter. The Department of Energy says data centers used about 4.4 percent of U.S. electricity in 2023. It also cites a projection that data centers could use about 6.7 to 12 percent by 2028.

That range does not prove a household bill increase. It proves the cost question is large enough to matter. A load that large needs power plants, transmission lines, substations, studies, backup plans, and rules for what happens when demand arrives faster than supply.

The pledge tries to answer the obvious question: who pays?

DOE’s current data-center resource hub says the pledge requires technology companies to build, bring, or buy new power, pay for required delivery upgrades, negotiate separate rates, invest in local jobs, and coordinate with grid operators. The company gets the new plug. The company pays for the new plug.

Under that rule, ordinary customers should not be used as the quiet backup wallet.

The promise works only if the new load carries the new cost.

FERC Is Asking The Same Question

On June 18, 2026, the Federal Energy Regulatory Commission ordered six regional grid operators to justify or reform the rules for data centers and other large energy users. FERC gave them 60 days to address large-load tariff rules.

One category was direct: preventing cost shifting and requiring transparency into transmission costs.

That phrase does the work. Cost shifting means one customer causes a cost and other customers pay it. Transparency means the public can see enough of the cost to know where it went.

FERC Commissioner Judy W. Chang made the warning sharper in a concurrence . She wrote that the Federal Power Act is a customer-protection statute. She also warned that voluntary commitments may fall short of the best protection, and that the assignment of power infrastructure costs sits at the center of the proceeding.

That makes the dispute a billing dispute.

If a company needs a new substation, the bill should know that. If a project needs a new transmission line, the rate case should know that. If a data center reserves power and then delays, the contract should know who pays for the stranded cost. If the answer is always “the general rate base,” then the public promise has failed its own premise.

A Real Example Shows Assignment

DOE’s Ohio fact sheet on the Portsmouth project gives one version of the promise in practice. DOE says SB Energy and AEP Ohio plan a large AI and power project, including $4.2 billion in transmission infrastructure, 10 gigawatts of new compute power, and 9.2 gigawatts of new natural-gas generation. DOE says SB Energy committed to paying for the new transmission infrastructure. DOE also says the project will use a dedicated data-center rate structure so the new power plants and transmission lines are paid by the project, not Ohio families or small businesses.

The useful part is the mechanism. It goes past the broad line “AI will help the economy.” It says who builds, who pays, and what kind of rate structure carries the cost.

The limit is clear. A fact sheet differs from a bill. A commitment differs from a completed tariff. A project announcement differs from a household meter.

The public record sits in contracts, filings, rate structures, and bills.

When cost assignment is unclear, the risk shifts toward ordinary bills.

The Ledger

If AI data centers are public goods, their backers can say so. They can argue that the country needs compute capacity, jobs, military strength, scientific power, and private investment. Some of that may be true.

If AI data centers are private loads, their backers must also say so. A private load should pay for the power system it requires.

Both claims can be true at once. A project can serve a national goal and also create a local cost. A company can build useful infrastructure and need firm rules about who pays. A pledge can point in the right direction and need public accounting before ordinary customers should trust it.

If the new load needs the new plug, the new load should carry the new bill.

Ratepayer protection has to be a ledger. Liking or fearing AI cannot settle it. The power plant has a cost. The transmission line has a cost. The substation has a cost. The reserved capacity has a cost. The delay risk has a cost.

The promise is real only when those costs do not disappear into everyone else’s bill.