Policymakers Increasingly Expect Data Centers to “Bring, Buy, or Build” Generation

Recent policy discussions concerning data center development have increasingly focused on requiring large load customers to bear the costs associated with serving that load. The concept has gained prominence through the White House's Ratepayer Protection Pledge, which calls on participating hyperscalers and AI companies to "build, bring, or buy" the generation resources needed to support their facilities and to pay for associated infrastructure upgrades.[1] Signatories also committed to bear those costs regardless of whether the resulting electric service is ultimately utilized.

The issue is particularly timely following the July 23, 2026 expansion of the White House's Ratepayer Protection Pledge to include utilities, electric cooperatives, governors, and additional data-center developers.[2]

At a high level, the Ratepayer Protection Pledge rests on a straightforward premise: large load customers should support the resources necessary to serve their own growth. Recent commentary surrounding the expanded pledge noted that utilities often take different positions regarding the extent to which customers may directly procure power or generation resources. As Jeff Dennis, executive director of the Electricity Customer Alliance, observed, implementation of the pledge's "bring, build, or buy" concept may prove challenging where utilities and customers have differing views regarding customer procurement of dedicated power supplies.[3]

The implementation challenges Dennis identified raise a fundamental question for developers and investors alike: what commercially viable pathways exist for data centers to obtain dedicated generation, particularly in vertically integrated utility states?

Why State Utility Law Matters

The emerging expectation that data centers will "bring, buy, or build" generation often assumes that developers can freely contract for dedicated power supplies if they are willing to bear the associated costs. In many vertically integrated utility states, however, that assumption is not always straightforward.

Large industrial customers have long utilized on-site generation, cogeneration, and other self-supply arrangements. Accordingly, state utility laws generally do not prohibit a customer from owning generation used to serve its own load. The challenge for many modern data-center projects is scale. While self-supply may be legally permissible, a dedicated 500 MW or larger generation facility can require billions of dollars of capital investment and may not align with a developer's preferred financing, ownership, or risk allocation strategy. As a result, developers frequently explore structures in which an independent power producer develops, owns, and operates the generation facility while supplying power to the data center under a long-term commercial arrangement.

In many vertically integrated states, retail electric service remains heavily regulated through exclusive service territories and utility-franchise frameworks. As a result, developers may discover relatively late in the planning process that a structure involving an independent power producer selling electricity directly to a co-located data center can raise questions regarding unauthorized retail electric service, even where the project involves a large behind-the-meter generating facility serving a data center campus. Whether a particular structure is permissible may depend on factors such as ownership of the generation facility, ownership of the load, whether electricity is sold or merely supplied within a common ownership structure, and whether the arrangement falls within a recognized statutory or regulatory exception. In many cases, those questions arise before a developer reaches the more familiar issues of permitting, fuel supply, transmission service, or interconnection.

Ownership structure therefore becomes critically important. A project that may be viewed as self-supply under one ownership arrangement may present very different regulatory considerations if generation and load are separately owned. As a result, issues such as common ownership and utility participation frequently become central to project design long before construction begins.  

Vertically integrated states frequently recognize limited exceptions that permit entities other than the incumbent utility to provide or supply electricity in certain circumstances. Examples may include landlord-tenant arrangements, customer-owned generation, renewable energy leasing programs, and other narrow statutory or regulatory exceptions. South Carolina, for example, expressly authorizes leasing arrangements for customer-sited renewable generation facilities and provides that a qualifying lessor is not treated as an electric utility when operating within the statutory framework.[4] Iowa courts have likewise permitted certain behind-the-meter solar power purchase arrangements without automatically treating the developer as a public utility.[5]

State approaches vary considerably. Many public utility statutes focus on whether a provider offers service "to the public" or "for the public," but states differ in how those concepts are interpreted. As a result, arrangements that are permissible in one jurisdiction may face very different treatment in another.

Many of these exceptions were developed for circumstances in which the underlying policy concerns differ significantly from those raised by a 500 MW hyperscale data center seeking dedicated generation. Whether concepts developed for customer-sited renewable facilities, landlord-tenant relationships, or traditional industrial projects readily translate to dedicated generation serving a large data center is often less clear. Arrangements that work in those contexts may not accommodate the scale, financing requirements, operational expectations, and ownership structures commonly associated with modern data-center development. Accordingly, the relevant legal question is often not whether a path exists but whether the available path aligns with the commercial objectives of the developer, investor, generation provider, and utility.

Commercial Viability Matters as Much as Legal Viability

The current debate surrounding data center load growth focuses on familiar questions: who should pay for new generation resources, who should fund grid upgrades, and how existing customers can be protected from subsidizing large new loads. Those questions are important, but they assume the existence of a commercially viable mechanism through which a data center can obtain dedicated power in the first place.

In many vertically integrated utility states, the more difficult question may not be whether generation can be developed, but whether developers can access that generation through a structure that aligns with their commercial objectives. A project may be technically feasible and legally permissible while still failing to satisfy key business objectives. Developers evaluating dedicated generation typically prioritize speed-to-power, financing flexibility, operational control, long-term power costs, risk allocation, and future expansion opportunities. As a result, the relevant inquiry is often not simply whether state law permits a particular arrangement, but whether the available structure produces a commercially attractive outcome.

Ownership structure frequently becomes central to that analysis. A structure that appears straightforward from a commercial or project-finance perspective may trigger very different regulatory considerations once electricity must cross an ownership boundary. Common ownership, affiliate ownership, utility-participation models, and utility-administered arrangements may produce materially different regulatory and commercial outcomes. In some jurisdictions, the lowest-risk regulatory pathway may involve a greater degree of utility ownership, utility participation, or utility administration than the developer originally anticipated. Accordingly, developers evaluating dedicated generation strategies should address utility law issues early in project development. Questions regarding ownership, power supply arrangements, and utility participation may prove as important as generation technology, fuel supply, permitting, or interconnection.

Conclusion

The emerging consensus that data centers should "bring, buy, or build" generation reflects a reasonable policy objective: customers driving significant new load growth should bear the costs associated with serving that load rather than shifting those costs to existing customers.

The inclusion of utilities, cooperatives, and governors in the July 2026 expansion of the Ratepayer Protection Pledge underscores that implementation may now become as important as policy. In some vertically integrated states, a central question will be how "bring, buy, or build" commitments can be implemented within existing utility service and retail electricity frameworks.

As the discussion shifts from policy objectives to implementation, a key issue may become whether state utility law provides commercially viable pathways through which data centers can obtain dedicated power while satisfying the objectives of developers, investors, utilities, regulators, and customers.

For developers considering dedicated generation strategies, that analysis should begin early. Before asking whether a project can bring, buy, or build generation, stakeholders should understand whether the applicable regulatory framework permits commercially viable ownership and procurement structures in the first place.

[1] Ratepayer Protection Pledge, Proclamation No. 11014, 91 Fed. Reg. 11,439 (Mar. 9, 2026).

[2] Mara Hoplamazian, Trump Expands His Ratepayer Pledge with Republican Governors, Utilities, NOTUS (July 23, 2026), https://www.notus.org/energy/trump-expands-ratepayer-pledge-republican-governors-utilities.

[3] Id.

[4] See S.C. Code Ann. § 58‑27‑2610.

[5] SZ Enterprises, LLC d/b/a Eagle Point Solar v. Iowa Utilities Board, 850 N.W.2d 441 (Iowa 2014).