Nicole Polisar

Associate Editor

Loyola University Chicago School of Law, JD 2027

Electric bills are rising in many places, and the rapid expansion of AI data centers is adding new pressure to the power system. The big issue is how the electric grid pays for the infrastructure needed to serve rapidly growing electricity demand tied to AI. Serving that demand can require costly upgrades to the electric grid as well as securing additional electricity supply. When those costs are recovered through broadly applied rates instead of being assigned to the large new loads that triggered them, residential customers can see higher bills. State commissions and federal regulators influence these outcomes through tariffs, cost-allocation rules, and market design. As AI electricity use accelerates, questions of fairness and reliability have moved to the forefront of energy regulation.

Why electric bills are rising

A common misconception is that electric bills are rising because data centers are not paying for the electricity they consume – this is not accurate. The real culprit is the cost of expanding the electric grid to keep up with the rapidly growing electricity demand caused by AI data centers. The electric grid is a power system that generates electricity and delivers that electricity to customers. When demand rises this quickly, utility companies have to make significant investments in building or upgrading the electric grid. Recent data helps explain why utilities anticipate new spending – The Berkeley Lab’s  Energy Usage Report concluded that the total U.S. data center electricity use tripled over the last decade and could nearly double—or triple—again by 2028.

That growth raises a basic question: who pays for the upgrades required to serve it? In this context, a large-load is a single customer whose electricity demand is so large at one location that serving that customer can require special planning and major grid upgrades. Data centers are a common example. A large-load project can trigger upgrades that benefit the new customer first, and the financing of those upgrades depends on how regulators allow utility companies to recover costs. When upgrade costs are recovered through increased general rates rather than charging the new large-load customers, residential customers may experience higher bills. At its core, the fairness debate turns on cost causation: customers who cause new costs should bear them.

How regulators are responding with large-load tariffs

At the state level, Public Utility Commissions (PUCs) set retail electric rates and review utility spending under authority aimed at protecting the public interest. At the federal level, the Federal Power Act (FPA) requires the Federal Energy Regulatory Commission (FERC) jurisdictional rates to be just and reasonable. It also authorizes FERC to oversee tariffs and practices affecting wholesale rates and transmission service. The Congressional Research Service (CRS) explains that FERC carries out that role primarily in FPA sections 205 and 206, which govern filings and complaints. These state and federal processes shape how costs from large new electricity loads are assigned across customers through tariffs, riders, and market rules.

One emerging compliance tool is the use of large-load tariffs. These tariffs impose specialized charges and contractual terms intended to prevent cross-subsidization. Ohio provides a clear example. The Public Utilities Commission of Ohio (PUCO) approved a data center tariff settlement in July 2025, mandating that large new data center customers must pay for a minimum share of subscribed usage even if they consume less, which helps ensure infrastructure costs are recovered from the customers driving the expansion. Analysts highlight additional protective mechanisms embedded in these tariffs, such as long-term contract commitments and exit fees. These protective mechanisms reduce the risk of stranded costs if a data center project is cancelled after upgrades are built. Stranded costs are expenses for infrastructure already paid for or constructed that no longer have the expected customer load to cover them, which can push those costs onto other ratepayers. Together, these provisions seek to align infrastructure investment with the entities that necessitate it.

What future compliance steps could reduce bill impacts and climate risk

Several policy tools could reduce the risk that AI-driven demand growth leads to higher household bills. Measures such as minimum bills, upfront contributions, and exit fees can reduce the likelihood that stranded costs are shifted to other customers. Another step is to require standardized reporting on large-load energy performance and resource use so regulators can evaluate mitigation plans using credible data. The European Union (EU) has already moved in this direction; under its revised Energy Efficiency Directive framework, the EU created a database to collect and publish information on data center energy performance. In the U.S., state commissions could pair large-load tariffs with transparency requirements around contracted demand, ramp-up schedules, and system upgrade triggers that drive costs. At the federal level, continued tariff clarification can reduce reliability risk and limit the incentive to overbuild infrastructure as a quick fix for uncertainty. Together, these steps aim to protect customers while keeping reliability and environmental impacts in view.