Data centers in the United States are expected to consume large amounts of electricity in the coming years.
However, frequent announcements of new projects alongside cancellations make it difficult to determine how many facilities will actually be built and how much power they will ultimately require.
This uncertainty presents a challenge for the electricity sector. State utility regulators must decide how to allocate the costs of generating and delivering power to these operations, often referred to as “large load centers.” The issue is complex, and states are testing different approaches, each with its own advantages, limitations and risks.
A New Kind Of Electricity Customer
Large electricity users are not new. For decades, industries such as textile manufacturing and oil refining have consumed enough power to rival small cities. Historically, their construction timelines aligned with the development of new electricity infrastructure. If a utility needed to build a new gas-fired power plant to serve a factory, both projects could proceed on similar schedules and be completed within two-and-a-half to three years, allowing costs to be recovered once service began.
Modern data centers consume comparable amounts of electricity but can be constructed in as little as nine to 12 months. To meet anticipated demand, utilities may need to begin building new generation capacity—such as gas-fired plants or solar facilities with battery storage—one or two years before a data center breaks ground.
During that period, rapid advances in computing technology, artificial intelligence and energy efficiency can significantly alter how much electricity a data center ultimately uses. These technological shifts, combined with planning and logistical challenges, create substantial uncertainty about future demand and make it difficult for utilities to determine how much capacity to build.
Managing Development Risk
Uncertainty carries financial consequences. A utility may build generation capacity in advance only to find that some or all of it is unnecessary. Alternatively, insufficient capacity could be available when a data center comes online, driving up electricity prices. In either case, costs must be borne by someone.
Those costs can fall on utilities, data center customers or other electricity customers. Utilities generally limit their own risk through state regulatory processes. Under these systems, utilities must seek approval from regulators before passing operating and capital expenses—such as power plants and grid upgrades—on to customers. Regulators evaluate whether the expenses are reasonable and necessary to provide reliable service.
Utilities therefore present evidence to justify investments linked to projected data center demand. Regulators are then left to determine how to fairly distribute those costs between data centers and other customers, including households and small businesses. Different states have adopted different solutions.
Kentucky’s Approach: Proving Need
Kentucky has attempted to address demand uncertainty by conditionally approving two new natural gas power plants. Utilities Louisville Gas & Electric and Kentucky Utilities must demonstrate that the plants will be needed and used. Given the long timelines involved, providing such proof is difficult.
A contract or agreement with a data center may be sufficient for regulatory approval. However, if the data center is never built or requires significantly less electricity than anticipated, utilities may be unable to recover costs from the company because billing is typically based on actual usage. In that situation, remaining customers could end up paying for unused capacity.
Ohio’s Demand Ratchet And Credit Guarantees
Ohio has implemented a more prescriptive framework. American Electric Power offers a special rate plan for data centers and other large users that includes a “demand ratchet.” Under this mechanism, a customer’s monthly bill is based on either current usage or 85% of the highest monthly demand over the previous 11 months, whichever is higher.
This structure protects against large swings in electricity use that would otherwise reduce payments, helping ensure data centers contribute consistently to the cost of maintaining adequate capacity. However, the ratchet effectively secures payments for only a year, which may be insufficient given the long-term nature of power plant investments.
Other states have gone further. Florida regulators have approved agreements requiring data centers to pay for 70% of contracted demand even if the electricity is not used.
Ohio also requires large customers to provide credit guarantees—such as deposits, letters of credit or parent-company guarantees—equal to 50% of their expected minimum bill. While this reduces risk for other customers, it raises concerns. Utilities may contract with subsidiaries created solely to operate individual data centers rather than with large, well-capitalized technology firms. If a subsidiary’s plans change or it declares bankruptcy, remaining customers could still bear the costs.
Using Flexibility To Reduce Risk
One advantage of data centers is their operational flexibility. In states such as Texas, data centers can earn revenue by adjusting electricity use in response to grid conditions. Sharing a portion of these earnings can help compensate other customers who assumed some of the investment risk.
Missouri has adopted a similar mechanism: when utilities earn additional revenue from large customers, 65% of that increase is returned to other ratepayers.
The U.S. electricity system is undergoing significant change, but the scale and timing remain uncertain. States are experimenting with different ways to distribute the costs associated with that uncertainty. Understanding the strengths and weaknesses of these approaches is critical to developing policies that are fair to utilities, large electricity users and the broader customer base.
Source: The Invading Sea