In April 2026, the Federal Energy Regulatory Commission finalized a rule that rewrote how large loads connect to the transmission grid. If you run interconnection studies, manage a load-growth forecast, or advise a data-center customer shopping for capacity, the FERC Large-Load Rule is already reshaping the work in front of you.

Why FERC Had to Act: The Grid Did Not Expect This

For most of the modern grid's history, load forecasters could rely on a reassuring truth: demand grew slowly, predictably, and in small increments tied to population, economic activity, and weather. Utilities built transmission planning models on that assumption and refined them over decades. The models were good precisely because reality was smooth.

Then data centers arrived at scale.

A hyperscale data center drawing 100 to 400 MW does not look like anything in a utility's historical load file. It does not ramp up with the morning commute. It does not correlate with air-conditioning load on a summer afternoon. It arrives as a step function: the facility interconnects, the meters spin, and the load curve has a new plateau it never had before. Traditional ARIMA and regression-based forecasting models, trained on decades of smooth growth, are not calibrated for step loads of this magnitude. The model has simply never seen one and has no idea what one looks like.

The interconnection queue reflects the same shock. By the end of 2025, more than 2,060 gigawatts of generation and storage capacity sat in queues across the country, with a median request-to-commercial-operation time that had roughly doubled to more than four years. A large fraction of those queue positions will withdraw. Meanwhile, a wave of large-load requests from AI data centers and other compute infrastructure was arriving alongside the generation queue, creating a policy vacuum: the rules written to connect generation projects did not cleanly address the special reliability questions raised by a 300 MW compute load that wants to connect to the transmission network.

FERC recognized that the existing framework was not built for this environment. In April 2026, FERC committed to issue its large-load interconnection rule by the end of June 2026 (Docket RM26-4-000), the first national reset of large-load interconnection policy in a generation.

What the Rule Actually Says: The 20 MW Threshold

The heart of the FERC Large-Load Rule is a threshold: loads seeking to interconnect at or above 20 megawatts on the transmission system now face a defined federal review process that did not previously exist in a uniform national form. Before this rule, large-load interconnection was handled patchwork-style: some ISOs/RTOs had their own large-load queue procedures, some utilities had distribution-level processes that were stretched to cover transmission-level requests, and some applicants navigated a gray zone of special agreements negotiated on a case-by-case basis.

The new rule imposes several key elements that every interconnection professional needs to understand:

A Standardized Application and Review Process

Applicants for loads above 20 MW must now go through a defined intake procedure that includes specific technical data requirements. Think of this as the load equivalent of an interconnection request for a generator: the applicant must provide operational load profiles, expected ramp characteristics, proposed curtailment capabilities, and backup supply arrangements. The days of submitting a letter of intent and a single-line diagram are over for these projects.

For utility interconnection engineers, this means your intake queue just changed. If you were processing large-load requests under an ad-hoc process, you now have a federal standard to meet. If you are using AI-assisted intake review tools to flag incomplete applications, those tools need to be updated to reflect the new required fields.

New Reliability Study Requirements

The rule requires that large loads above the threshold be studied for their impact on transmission reliability, including N-1 contingency analysis. This is the study methodology that grid planners know well from generation interconnection: the system must remain secure even if the largest single element fails. Applying N-1 thinking to a large load is a meaningful shift, because a 300 MW data center that trips suddenly represents a load loss event that can affect grid frequency and voltage stability across a wide area.

Notably, the rule opens the door to requiring large loads to provide some form of curtailment capability as a condition of interconnection, similar to how generation projects are required to provide ride-through and reactive support. This is where the FERC rule connects directly to NERC's Computational Load Entity work, which we cover in the next lesson.

A Framework for Cost Allocation

One of the thorniest issues in large-load interconnection is who pays for the network upgrades needed to accommodate the new customer. Under the historic generation interconnection model, the generator typically pays for the network upgrades it triggers ("participant funding"). The rule establishes a parallel framework for large loads, making clear that a data center seeking a 400 MW interconnection cannot simply assume that the existing network will absorb the request at no cost to the applicant.

For a utility rate-case professional, this matters because the cost-allocation decision will eventually show up as a proposed revenue requirement. Understanding whether upgrade costs are socialized across ratepayers or allocated to the large load is a central question in how your utility structures the large-load tariff you will have to defend before your state commission.

What It Means for Your Utility Right Now

Let's translate this from FERC Order language to operational reality for three common roles.

The Interconnection Engineer

Your study queue now includes a new category of applicant with federal procedural rights. If a data center company files a large-load interconnection request above 20 MW, you cannot process it under an informal process that varies case by case. You need a defined intake checklist, a documented study methodology, and a timeline that complies with the rule's procedural deadlines. AI-assisted study intake tools that were built before mid-2026 may not yet reflect the new required data fields. Before using one of those tools to triage an application, verify that its checklist has been updated.

The Load Forecaster

The rule creates a new data trail: large-load applicants must now submit operational load profiles. That is forecast gold. A 300 MW data center that files with an expected load profile of 250 to 320 MW, 24 hours a day, 365 days a year, gives your forecasting team something far more concrete than the historical proxy approach of "assume the facility ramps to nameplate over 18 months and stay there." Build a workflow to ingest these filed profiles into your load forecasting pipeline. Treat them as conditional scenarios until the facility actually interconnects, then treat them as committed load once they reach commercial operation.

The step-load risk is real: a 300 MW facility interconnecting in Q3 of a planning year can push your annual peak forecast well outside the confidence band of any model trained only on historical data. The rule's data requirements are your early warning system if you use them.

The Regulatory Affairs Professional

The rule creates a new FERC-level procedural framework, but it does not preempt state commission authority over how a utility structures its large-load tariff, how it recovers upgrade costs, or how it classifies a large-load customer for rate purposes. You will need to watch for your ISO/RTO's compliance filing to FERC, which will implement the rule's requirements in the regional tariff. You will also need to coordinate with your state commission as the large-load tariff questions inevitably migrate from FERC dockets to state rate cases. The two processes will run in parallel, and the testimony you file at the state level will need to be consistent with the positions your utility took in FERC proceedings.

Worked Example: The 400 MW Campus Request

A large technology company files a large-load interconnection request for a new data center campus in your service territory. The stated load is 400 MW, with a load factor expected above 90 percent and minimal flexible curtailment capability in the first three years of operation. Here is how the pre-rule and post-rule worlds differ.

Before the rule, your utility might have responded with a feasibility study scoped under your existing generation-side interconnection procedures, modified on the fly to address load-side questions your process was not designed to handle. The applicant's legal team might have argued that certain upgrade costs should be socialized. Your planning team might have absorbed the load profile as a planning assumption without any formal data disclosure requirement. The whole process could take years, and the outcome would depend heavily on how your ISO/RTO decided to treat the request.

After the rule, the applicant submits a standardized application with the required technical data, including the load profile, ramp characteristics, and proposed curtailment commitments. Your interconnection team runs the process under a defined timeline with federally specified procedural rights for the applicant. The N-1 reliability study is a required deliverable, not a judgment call. The cost-allocation framework tells both parties which upgrades the applicant will pay for and which costs, if any, may be socialized.

Where AI tools help in this scenario: an AI-assisted review of the applicant's load profile against your system capacity model can flag potential N-1 violations before the full study is complete, giving your engineering team early warning of the most expensive upgrade triggers. An AI-assisted drafting tool can help produce the first version of the interconnection agreement boilerplate. Both outputs require human engineer sign-off before they drive any decision. The model speeds the throughput; the engineer owns the result.

The FERC Large-Load Rule is not a technology rule. It is a reliability rule that happens to be triggered by technology. Understanding it as a reliability standard, not a data-center accommodation, is the frame that will serve you best in every proceeding where it appears.

The Connection to the AI Forecasting Challenge

Peak demand is forecast to grow by approximately 166 gigawatts over the next five years, with roughly 90 gigawatts of that growth attributable to data centers. Data-center electricity consumption is projected to rise from 176 TWh in 2023 to somewhere between 325 and 580 TWh by 2028. Those are numbers that belong in every load forecaster's opening slide, not because you should cite them verbatim in a rate case, but because they establish the order of magnitude of the challenge.

No statistical forecasting model trained purely on historical utility data anticipated a load-growth trajectory of this shape. The FERC rule's data disclosure requirements for large-load applicants are one of the few mechanisms that give utilities advance notice of where the step loads are coming and approximately how large they will be. Treat the large-load application pipeline as a leading indicator for your five-year forecast, not as a legal process you hand off to interconnection engineers and forget.

AI forecasting tools that incorporate pipeline data, operational profiles from large-load applicants, and satellite or permitting signals for data-center construction can achieve meaningfully better accuracy on the five-year forward curve than models that rely solely on historical consumption data. The accuracy claim often cited is approximately 1 to 2 percent mean absolute percentage error (MAPE) at the day-ahead horizon for AI-based models, compared to 3 to 5 percent for traditional ARIMA or regression approaches. That gap widens substantially on planning-horizon forecasts when large step loads are present. Verify those numbers against your own system before putting them in testimony, but understand the directional logic: better input data, better model, better forecast.

Open Questions and What to Watch

Several questions remain unresolved as the rule's implementation unfolds through ISO/RTO compliance filings and state proceedings:

• Curtailment requirements: The rule opened the door to requiring large loads to offer curtailment capability as a condition of interconnection. Whether and how ISOs/RTOs implement curtailment requirements in their compliance filings will determine whether data centers become reliability assets (they can reduce load when the grid is stressed) or reliability risks (they run at full capacity regardless of grid conditions).
• Interaction with the NERC Computational Load Entity standard: The FERC rule establishes the interconnection process; the NERC CLE standard (committed for December 31, 2026) establishes ongoing reliability obligations for large compute loads. Understanding how these two frameworks interact is a compliance-lead priority for any utility with data-center customers above the threshold.
• State commission implementation: FERC regulates transmission interconnection. State commissions regulate distribution interconnection and retail tariffs. A data center that connects at the transmission level may still have distribution-level interactions at a substation your utility owns. The boundary questions between FERC and state jurisdiction are not fully resolved, and you should expect litigation over them.
• Queue management for co-located requests: Some large loads want to co-locate with a generation resource (a data center plus behind-the-meter solar, for example). The rule does not fully address co-location scenarios, and the queue management questions will be worked out in individual ISO/RTO compliance filings.

The practical reality is that large-load interconnection requests are already in your queue, and the FERC rule's procedural requirements apply now. Utilities that treat the rule as a future compliance item will find themselves processing applications under a standard they have not yet built processes for, which means delays, data requests to applicants that are not clearly grounded in the rule, and reliability studies that do not fully satisfy the new requirements.

The preparation checklist is straightforward. Update your intake form to include the new required data fields: operational load profile, ramp characteristics, proposed curtailment capability, and backup supply arrangement. Train your interconnection engineers on N-1 contingency analysis for load events, which differs from the generator trip scenario they typically analyze. Establish a coordination workflow with your ISO/RTO so that large-load applications requiring transmission-level studies flow to the right desk on the right timeline. Designate a rate-case coordinator who tracks FERC large-load filings and flags those with potential cost-allocation implications for your state commission proceedings. These steps take weeks, not months, and they determine whether your utility leads or lags in a regulatory environment that is moving faster than the traditional utility planning cycle.

Key Takeaways

• FERC committed in April 2026 to issue its large-load interconnection rule by the end of June 2026 (Docket RM26-4-000), establishing the first uniform national framework for interconnecting loads above 20 MW, driven primarily by AI data-center demand.
• The rule requires standardized applications with technical data, N-1 reliability studies, and a defined cost-allocation framework for network upgrades, ending the era of ad-hoc large-load processes.
• Interconnection engineers need updated intake checklists and study methodologies that reflect the new federal requirements; AI-assisted review tools must be checked for compliance with the new data fields.
• Load forecasters should treat the large-load application pipeline as a leading indicator for five-year peak forecasts, using filed load profiles as conditional scenarios until interconnection is achieved.
• The rule does not preempt state commission authority over tariffs, cost recovery, or retail classification; regulatory affairs professionals must coordinate FERC and state proceedings simultaneously.
• Open questions around curtailment obligations, NERC CLE coordination, co-location, and state jurisdiction will be resolved through ISO/RTO compliance filings and litigation over the next 12 to 24 months.
• AI tools can accelerate intake review and study drafting, but every reliability and cost-allocation output requires engineer sign-off; the rule, not the tool, owns the obligation.