The first two alerts in this series traced the Federal Energy Regulatory Commission’s (FERC or the Commission) regulatory trajectory and the political economy now shaping data center power. This alert turns to the structuring question that sits at the center of every behind-the-meter generation and co-located load transaction serving data centers (collectively, BTM): how to design an arrangement that achieves the developer’s commercial objectives while managing, or, where the project configuration permits, avoiding FERC jurisdiction.
Structure determines jurisdiction. That is the single most important proposition in this area of practice. The physical configuration of the generation-to-load connection, the ownership architecture, the contractual relationship between the generator and the data center, and the presence or absence of a grid interconnection collectively determine whether the arrangement falls within FERC’s regulatory reach or outside it. A transaction that is structured carefully can avoid federal wholesale jurisdiction entirely. A transaction that inadvertently triggers FERC jurisdiction can expose the generator to market-based rate (MBR) authorization requirements, ongoing compliance and reporting obligations, potential refund liability, and, in PJM Interconnection (PJM) and the Southwest Power Pool (SPP), the full weight of the co-location frameworks described in Alert 1. Those consequences are retroactive, costly, and difficult to unwind.
This alert examines the principal structural choices, identifies the jurisdictional triggers associated with each, and provides practical guidance for developers, sponsors, lenders, and their counsel.
The Two Jurisdictional Triggers
FERC jurisdiction over electricity transactions attaches through two independent pathways under Section 201(b)(1) of the Federal Power Act (FPA). Either trigger, standing alone, is sufficient to bring the arrangement within FERC’s regulatory reach.
The first is the sale-for-resale trigger. The FPA grants FERC jurisdiction over “the sale of electric energy at wholesale in interstate commerce,” and defines a wholesale sale as “a sale of electric energy to any person for resale.” The critical distinction lies between sales (which trigger jurisdiction when they are for resale) and consumption (which does not). If the data center is the end consumer of the electricity and does not resell it, the transaction is a retail sale reserved to state jurisdiction under FPA Section 201(b)(1). Self-supply by a single entity that generates and consumes its own power is the clearest case: no sale occurs at all, and FERC jurisdiction does not attach.
This distinction proves more complex in practice than in theory. Three increasingly common data center power arrangements illustrate the spectrum. This alert discusses each in turn below.
The second trigger is the transmission nexus. The FPA grants FERC authority over “the transmission of electric energy in interstate commerce” and “all facilities for such transmission or sale.” This transmission jurisdiction extends broadly. If generation facilities connect to the grid, even through a backup or supplemental interconnection, that connection may bring the arrangement within FERC’s reach regardless of whether the underlying sale is retail or wholesale. Courts and FERC have recognized limited exceptions for “radial” interconnections that are sole-use, limited and discrete, radial in nature, and not part of an integrated transmission network. But the exception is narrow, fact-specific, and cannot be assumed without careful engineering and legal analysis.
Three Structures, Three Risk Profiles
Consider three increasingly common arrangements for a 200 MW natural gas plant adjacent to a data center facility.
Structure 1: Single-Entity Self-Supply. DataCo owns and operates both the data center and the 200 MW plant. The plant’s output flows entirely to the data center via a dedicated interconnection with no path to the grid. No electricity crosses the boundary of a single legal entity, and no sale occurs. FERC jurisdiction does not attach under either the sale-for-resale trigger or the transmission nexus (assuming no grid interconnection). This is the cleanest structure from a jurisdictional perspective. Its limitation is that it requires the data center operator to be in the power generation business, including managing fuel procurement, environmental permitting, plant operations, and maintenance, functions that many data center companies prefer to outsource.
Structure 2: Third-Party PPA. GenCo builds and operates the 200 MW plant. GenCo sells 100% of the plant’s output to DataCo under a 20-year power purchase agreement (PPA). The power flows through a dedicated interconnection with no grid connection. DataCo consumes all of the electricity and does not resell any portion. This arrangement should constitute a retail sale outside FERC’s wholesale jurisdiction, because DataCo is the end consumer and does not purchase for resale. However, FERC has long maintained that corporate separateness alone does not determine jurisdictional questions. The Commission looks to the substance of an arrangement, not merely its form, and considers whether the overall structure and purpose of the arrangement constitute a wholesale transaction regardless of the label applied by the parties. If DataCo were to resell any portion of the output, or if surplus generation were to flow to the grid through an interconnection, the sale-for-resale trigger could be implicated. And if the dedicated interconnection between GenCo’s plant and DataCo’s facility were synchronized with the grid, the transmission nexus could be triggered independently.
Structure 3: Affiliate Wholesale. GenCo and DataCo are corporate affiliates. GenCo sells the plant’s output into the wholesale market, while DataCo purchases power from the grid at retail rates through the local utility. This structure clearly involves wholesale sales requiring FERC MBR authorization and, in organized markets administered by regional transmission organizations (RTOs), full participation in the applicable RTO’s capacity, energy, and ancillary service frameworks. It provides no BTM benefit and no queue avoidance. It is included here to mark the outer boundary of the jurisdictional spectrum.
Structure 1 avoids FERC jurisdiction entirely. Structure 3 is fully within it. Structure 2 occupies the gray zone where most live transactions sit and where structuring precision determines the outcome.
Managing the Gray Zone: Structural Choices That Reduce Jurisdictional Exposure
For developers pursuing Structure 2 arrangements (or variations of it), several structural choices may reduce the risk of FERC jurisdictional exposure. None of these choices provides certainty in the absence of a FERC declaratory order or MBR filing, but each shifts the analysis in a favorable direction.
Ownership Integration. The strongest defense against the sale-for-resale trigger is eliminating the “sale” altogether. If GenCo and DataCo form a single-purpose limited liability company (LLC) that owns both the generation facility and the data center, the transfer of electricity from the plant to the load is an internal allocation within a single entity, not a sale between separate parties. The LLC structure should have genuine economic substance: shared capital contributions, integrated governance, common operational control, and a business purpose beyond jurisdictional avoidance. FERC’s substance-over-form analysis means that a structure that is formally integrated but functionally operates as a bilateral PPA between two independent parties may not withstand scrutiny. The more genuine the integration, the stronger the jurisdictional position.
For sponsors and their counsel, the ownership integration approach raises practical considerations that deserve attention at the term sheet stage. An integrated LLC typically means the sponsor and the data center customer (or their respective affiliates) are co-owners of the generation asset. That affects the capital stack, the allocation of development risk, the tax treatment (including whether the sponsor can claim depreciation and energy tax credits), and the exit mechanics. If the sponsor’s investment thesis contemplates selling the generation asset independently of the data center, or if the data center customer’s business model does not include co-ownership of generation, the integrated LLC may not be commercially feasible, and the parties may need to rely on one of the alternative structural approaches described below.
Deep Operational Integration. Where separate legal entities are required (for tax, financing, liability, or commercial reasons), the arrangement should demonstrate operational integration sufficient for FERC to treat the electricity transfer as an internal allocation rather than a wholesale sale. Indicators that FERC has considered in evaluating whether arrangements between affiliated entities constitute wholesale sales include common control of dispatch and operations, shared planning and investment decisions, integrated fuel procurement, absence of arm’s-length price negotiation, and whether the entities function as a single economic enterprise even though they are legally separate. The line between “deep operational integration” (which may support a characterization as internal transfer) and “separate entities with a PPA” (which constitutes a sale requiring jurisdictional analysis) is not bright, and FERC has provided limited guidance on where it falls. Developers pursuing this approach should document the operational integration comprehensively, because the analytical record will be important if FERC or an intervenor challenges the arrangement.
Prophylactic MBR Authorization. Where any material jurisdictional ambiguity exists, obtaining MBR authorization from FERC as a precautionary measure may be the most prudent path. MBR authorization requires demonstrating a lack of horizontal market power (the seller does not have the ability to raise prices above competitive levels) and vertical market power (the seller does not have the ability to erect barriers to entry through control of transmission). It also imposes ongoing compliance obligations, including quarterly transaction reporting, change-in-status filings, and triennial market power updates. These obligations are not trivial. But the alternative, operating without authorization and facing potential refund liability and enforcement action if FERC later determines that the arrangement constitutes a wholesale sale, presents a significantly less attractive risk profile. For lenders and equity sponsors, the existence of MBR authorization (even if never used operationally) may provide a level of regulatory comfort that facilitates financing on more favorable terms.
FERC Declaratory Orders. FERC’s declaratory order process under Rule 207 of the Commission’s Rules of Practice and Procedure allows developers to obtain jurisdictional clarity before commencing operations. The process typically takes six to 12 months and requires a detailed factual submission describing the arrangement, the parties, the physical configuration, and the basis for the requested jurisdictional determination. Declaratory orders are not binding on future Commissions in the same way that a rulemaking would be, but they provide significant comfort and reliance protection for the specific arrangement described. For large-capital projects where jurisdictional uncertainty affects financing terms, insurance requirements, or counterparty willingness to commit, the investment in a declaratory order proceeding may be justified.
The Physical Configuration: Islanded vs. Grid-Connected
The contractual structure and the physical configuration of the generation-to-load connection operate as independent jurisdictional variables. Even a contractually clean self-supply arrangement can trigger FERC transmission jurisdiction if the facilities are interconnected with the grid.
A facility that is electrically islanded from the grid, with no synchronization, no backup interconnection, and no path for power to flow to or from the interstate transmission system, presents the strongest case for avoiding FERC transmission jurisdiction. The generation facility delivers power to the data center through a dedicated line that does not touch the grid. No transmission of electric energy in interstate commerce occurs, and neither the generation facility nor the interconnecting line constitutes a facility used for such transmission.
The tradeoff for islanding is reliability. An islanded facility must be sized to meet the data center’s full load requirement plus sufficient reserve margin to cover maintenance outages and unplanned trips, which typically means overbuilding generation capacity by 15 to 30 percent relative to expected average consumption. That excess capacity represents significant capital investment that sits idle during normal operations. For a 500 MW data center, the incremental cost of 75 to 150 MW of reserve generation capacity can amount to hundreds of millions of dollars in additional capital expenditure. Sponsors and their financial advisors may wish to model this overbuilding cost against the regulatory cost savings from FERC avoidance to determine whether the islanded configuration produces a net economic advantage for the specific project.
Grid interconnection eliminates the overbuilding requirement by providing access to utility backup power during outages and maintenance. It may also enable the monetization of surplus generation through grid sales rather than curtailment. But any grid interconnection in a FERC-jurisdictional market subjects the facility to the applicable RTO’s interconnection procedures, study requirements, and, under the PJM Order (described in Alert 1), the new transmission service framework for co-located load. The choice between islanding and grid connection is therefore both an engineering decision and a regulatory decision, and it has direct implications for the project’s capital structure, operating economics, and bankability.
For lenders evaluating BTM generation projects, the islanded configuration presents a distinctive risk profile. The facility has no grid backup, meaning that any generation outage directly affects the data center’s operations and revenue. Lenders may require higher debt-service coverage ratios, larger reserve accounts, or more comprehensive insurance coverage for islanded projects than for grid-connected projects. The generation technology, the fuel supply arrangements (including firm gas transportation and on-site storage), and the maintenance program become critical credit considerations in the absence of grid backup. Conversely, the islanded facility carries no risk of unanticipated transmission charges, regulatory reclassification, or RTO tariff changes, which may simplify the regulatory risk analysis in the credit memorandum.
The Phased Approach
For developers who need both speed to market and eventual grid access, a phased strategy may offer the best path to both objectives. The concept is straightforward: begin operations on a fully islanded basis under a self-supply or integrated LLC structure, which allows the data center to reach commercial operation and begin generating revenue without waiting for interconnection queue processing. Pursue grid interconnection studies and approvals in parallel. Transition to a grid-connected configuration, with the attendant transmission service and cost-sharing obligations, once interconnection rights are secured.
The phased approach has meaningful practical advantages. It compresses the time to revenue, which can be critical for sponsor internal rate of return and for meeting contractual delivery commitments to the data center customer. It provides a period of operational experience during the islanded phase that can inform the design of the grid-connected phase. And it allows the developer to observe how the regulatory framework (including the DOE Rulemaking Proposal, the PJM compliance filings, and the SPP High Impact Large Load (HILL) framework, each described in Alert 1) develops before committing to a specific transmission service election.
The execution is more complex than the concept. The regulatory analysis for the islanded phase differs from the grid-connected phase in several material respects, and both phases must be structured from the outset to accommodate the transition. The PPA or LLC agreement should contemplate the shift from islanded to grid-connected service, including triggers for when the transition occurs, allocation of the incremental transmission charges and study costs associated with grid connection, adjustment of the capacity commitment and pricing structure to reflect the availability of grid backup, and modification of the risk allocation (including force majeure, curtailment, and regulatory change provisions) to account for the RTO framework that takes effect upon interconnection. The interconnection application should be filed early enough in the islanded phase to allow the queue position and study process to mature by the time the developer is ready to transition. And the financing documents should address the transition as a planned event with defined conditions, not as a material change that triggers covenant defaults or renegotiation.
Lenders may view the phased approach favorably if the transition mechanics are well-defined, because it combines the regulatory simplicity of the islanded phase with the reliability and revenue advantages of grid connection. But the transition introduces its own risks: the grid connection may be delayed by queue backlogs, study costs may exceed initial estimates, and the RTO tariff provisions that apply at the time of interconnection may differ from those in effect when the project was initially structured. These risks should be addressed in the financing documents through conditions to the transition, cost caps or allocation mechanisms for interconnection study costs, and regulatory change provisions tied to the specific RTO framework. The data center customer’s service level agreement or offtake arrangement should also address the transition explicitly, including any planned outage window during the switchover from islanded to grid-connected service, the allocation of curtailment risk during the transition period, and whether the availability guarantee and liquidated damages framework changes upon grid connection to reflect the improved reliability profile.
The Landlord-Tenant Alternative
In certain states, a landlord-tenant utility exemption provides an additional structural pathway. Under this model, the generation owner holds both the generation facility and the real property, leases the site to the data center operator, and delivers power as a bundled component of the lease rather than as a separately metered and billed commodity. If structured properly, no “sale” of electricity occurs under applicable state utility law, the power delivery is a lease service, and the landlord is not classified as a utility.
The landlord-tenant structure has notable advantages for sponsors. The generation owner retains tax and GAAP ownership of the generation assets, preserving depreciation, bonus depreciation, and eligibility for energy tax credits (including credits under Section 45Y and Section 48E of the Inflation Reduction Act of 2022 (IRA), where the generation technology qualifies). The data center customer holds a lease interest rather than an ownership or leasehold interest in the generation assets, which simplifies the customer’s balance sheet treatment and avoids the complexities of asset lease classification for accounting purposes (specifically ASC 842).
The structure requires careful attention to several elements. The lease should include meaningful non-electricity terms (site access, infrastructure, maintenance obligations, shared facilities) so that it reads as a real property lease with bundled services rather than a power purchase agreement with a lease wrapper. The rent structure should avoid pure per-kWh volumetric charges that function as an electricity rate. A two-part structure consisting of a fixed capacity component and a variable operating cost pass-through may be more defensible, though the precise boundary between permissible cost recovery and impermissible retail sale will depend on the applicable state’s statutory framework and regulatory precedent. The lease term and renewal provisions should be consistent with commercial real property practice. And the arrangement should not involve separate metering or billing of electricity as a standalone commodity, which in several states has been identified by courts or regulatory commissions as the line between a bundled lease service and a regulated retail sale.
The landlord-tenant exemption is not available or has not been tested in every jurisdiction. The states where it has the strongest statutory or judicial support (and where it is most relevant for data center development at the scale contemplated in this series) are addressed in Alert 6.
From a FERC perspective, the landlord-tenant structure is most effective when combined with an islanded configuration. If there is no sale for resale (because the transaction is a lease service, not a wholesale sale) and no transmission in interstate commerce (because the facility is islanded), both jurisdictional triggers are avoided.
The Utility’s Likely Response
Developers should expect that incumbent utilities may challenge BTM generation arrangements, regardless of the structure chosen. Utilities have legitimate concerns about lost revenue, stranded cost recovery, and the impact of load departure on the remaining customer base. They also have an economic incentive to serve large loads rather than lose them to self-supply.
The most common challenges take several forms. A utility may file a complaint before the state public utility commission alleging that the arrangement constitutes an unauthorized retail sale or violates the utility’s certificated service territory. A utility may seek to impose prohibitive standby or backup rates on self-generating customers, effectively rendering BTM generation uneconomic. A utility may intervene in FERC proceedings or interconnection processes to raise cost allocation, reliability, or jurisdictional objections. And a utility may engage the state legislature or regulatory commission to seek changes in the statutory or regulatory framework that would restrict or increase the cost of BTM generation.
None of these challenges is necessarily fatal to a well-structured arrangement. State law in most jurisdictions protects the right of customers to self-generate, and standby rates are subject to regulatory review for reasonableness. But each challenge takes time and money to litigate, and the risk of an adverse outcome, even if the legal merits ultimately favor the developer, should be factored into the project timeline and budget. Developers who engage with the incumbent utility early in the process, who structure standby and backup service arrangements that address the utility’s cost recovery concerns, and who demonstrate willingness to participate in grid support (demand response, emergency generation, reliability coordination) may find less adversarial regulatory proceedings. The constructive engagement principle described in Alert 2 applies here with particular force.
PURPA: Limited Utility at Scale
The Public Utility Regulatory Policies Act of 1978 (PURPA) offers certain exemptions from FPA regulation for qualifying facilities (QFs), including exemption from FPA utility registration requirements, exemption from state utility rate regulation in certain circumstances, and mandatory purchase obligations from utilities at avoided cost rates. For data center BTM generation, QF status offers potential regulatory shelter.
However, PURPA’s constraints make it generally impractical for large-scale data center generation. The small power production facility pathway (covering renewables) caps at 80 MW, well below the 200 to 500 MW scale most developers contemplate. The cogeneration pathway has no size cap, which is its principal advantage, but it requires the facility to meet FERC’s operating and efficiency standards under 18 C.F.R. Part 292, including meaningful, useful thermal output in addition to electricity. Most data center loads are electricity-only. Using waste heat from generation for data center cooling is a plausible path to qualification, but the thermal output must be “useful” under FERC’s standards, not merely dissipated, and the facility must meet minimum efficiency thresholds. Whether waste heat used for cooling satisfies FERC’s useful thermal output requirement involves technical and regulatory uncertainties that are difficult to resolve as a planning assumption.
PURPA’s QF exemptions are most valuable when paired with third-party power sales structures, where federal and state utility regulation would otherwise apply most directly. Under self-supply or landlord-tenant structures, the state statutory exemptions typically do most of the regulatory work, and the incremental value of QF certification is limited to the mandatory purchase obligation (providing a floor for surplus output sales) and the Public Utility Holding Company Act (PUHCA) exemption (relevant only if the ownership structure involves a holding company with utility affiliates). These benefits should be weighed against the ongoing compliance burden of maintaining QF status.
For most large-scale data center BTM generation projects, PURPA exemptions are unlikely to serve as the primary regulatory strategy. They may assist smaller distributed generation projects, specialized cogeneration configurations where the data center’s cooling load can satisfy the thermal output requirements, or projects where the mandatory purchase obligation provides meaningful surplus sales revenue. But at the 200 MW and above scale, the structuring approaches described earlier in this alert provide a more reliable regulatory foundation.
Observations
Three principles emerge from this analysis.
First, the jurisdictional outcome is determined by the interaction of the contractual structure and the physical configuration. A self-supply or integrated LLC structure combined with an islanded configuration avoids both FERC jurisdictional triggers. A third-party PPA combined with grid interconnection potentially triggers both. Every other combination falls somewhere between these poles, and the regulatory risk profile of each must be evaluated on its specific facts.
Second, the optimal structure is not the one that minimizes regulatory exposure in isolation, but the one that best balances regulatory considerations against commercial, tax, financing, and operational objectives. A fully islanded self-supply arrangement is jurisdictionally clean but commercially restrictive. A third-party PPA with grid backup is jurisdictionally complex but commercially flexible. The right answer for a given project depends on the sponsor’s investment thesis, the data center customer’s operating requirements, the availability of financing for the chosen structure, and the state-specific regulatory framework.
Third, the regulatory landscape described in Alerts 1 and 2 is actively tightening the gray zone. The PJM Order, the SPP HILL framework, and the DOE Rulemaking Proposal (each described in Alert 1) are designed to bring co-located and BTM arrangements within a defined regulatory framework. Arrangements that rely on jurisdictional ambiguity, that depend on the absence of clear rules to avoid compliance, are increasingly exposed. The window for structuring around undefined tariff provisions is narrowing. Developers may wish to structure transactions on the assumption that the framework will be more defined, not less, by the time the project reaches commercial operation.
This is the third in a series of seven alerts examining the regulatory frameworks applicable to data center power across multiple jurisdictions. The next alert examines the Texas/ERCOT framework, the jurisdiction where these structuring choices are most straightforward, and the regulatory advantages relative to FERC-jurisdictional markets are most pronounced.
This alert is intended to provide a general overview of the structuring considerations applicable to behind-the-meter generation serving data center loads. It does not constitute legal advice, and the appropriate approach will depend on the specific facts, jurisdiction, and circumstances applicable to each project.
RJ Colwell is a Senior Associate at Davis Graham & Stubbs LLP in the Energy & Mining Group. His practice focuses on the regulatory, transactional, and structuring dimensions of data center power, advising data center developers, power generation companies, and their investors and lenders on FERC regulatory matters, behind-the-meter generation, co-located load arrangements, and energy infrastructure transactions. RJ can be reached at rj.colwell@davisgraham.com.