On March 15, 2026, NERC filed a commitment with FERC that will change how large compute facilities interact with the bulk electric system: the Computational Load Entity (CLE) registry. For the first time in the history of North American reliability standards, a class of electricity consumers will be treated as registered grid actors with enforceable obligations, not merely as loads to be served. If your utility hosts, plans for, or interconnects a data center or AI compute campus of any meaningful size, this is the most important regulatory development you will study in this program.

Why the Grid Needed a New Category

For most of electric utility history, the question of how to handle large industrial loads was settled by tariff rate schedules, interconnection agreements, and local distribution practice. The load paid its bill, the utility served the load, and NERC focused its reliability standards on the generators, transmission owners, and system operators who controlled real-time power flows. Consumers were, in regulatory terms, price-takers and demand objects. The grid was built around the assumption that the largest loads grew slowly, were geographically predictable, and had little ability to destabilize a transmission system on their own.

Data centers broke every one of those assumptions. A hyperscale AI training campus can absorb 500 megawatts at a single point of interconnection. It can ramp that load to full draw in seconds. It operates continuously, twenty-four hours a day, seven days a week, with a power usage effectiveness ratio designed to minimize wasted heat, not to accommodate grid conditions. And it locates wherever land, fiber, water, and power are available, meaning its arrival on a transmission circuit can be sudden and unforecast. When NERC planners modeled 2025 and 2026 summer peaks, they found that data-center growth in certain regions was exceeding any planning assumption derived from historical trends. The ARIMA models that utilities had used for thirty years to forecast load growth were producing outputs that bore no resemblance to actual meter reads on circuits serving compute campuses.

The May 2026 Level 3 Alert that NERC issued was the formal acknowledgment of this problem at the standards body level. A Level 3 Alert, for readers who are not NERC compliance leads, is the second-highest urgency tier in NERC's alert system. Level 1 and Level 2 Alerts are informational and advisory. Level 3 Alerts carry an expectation that registered entities will respond with concrete action. The May 2026 alert specifically flagged the reliability risks associated with large computational loads that were not registered, not obligated to provide operational transparency, and not required to participate in any demand-response or curtailment program that a transmission operator could invoke under stress conditions.

The CLE registry is NERC's structural answer to that alert. Rather than issuing another voluntary guidance document, NERC committed to creating a new registration category that would bring large compute loads inside the reliability tent. The commitment was made in the March 2026 FERC filing, with a delivery date of December 31, 2026.

What "Registered Grid Actor" Actually Means

The phrase "registered grid actor" carries specific weight in the NERC framework. When an entity is registered, it enters a compliance relationship with NERC and with its Regional Entity (RE). This relationship has five practical consequences that distinguish registration from simply being a large load customer on a utility tariff.

First, the entity must provide operational data. Registered entities are obligated to share real-time and planning data that the system operator and reliability coordinator need to model the bulk electric system accurately. For a CLE, this means metered demand, load forecasts, planned maintenance windows, and the parameters of any flexibility that the entity can offer. Before CLE registration, a data center had no such obligation. The utility could estimate its load from interval meter data, but the data center itself had no duty to report its operational plans to anyone beyond its contracted interconnecting utility.

Second, registered entities must comply with applicable reliability standards. NERC's standards library has more than one hundred active standards covering everything from transmission planning to emergency operations to cybersecurity. Not every standard applies to every registration category. The CLE category will carry a specific set of applicable standards, still to be defined through NERC's Standards Development Process (SDP), but the expectation from the March 2026 filing is that CLEs will be subject to standards covering demand response participation, load forecasting accuracy, emergency curtailment, and operational transparency.

Third, registration creates an audit and enforcement relationship. Regional Entities conduct compliance audits of registered entities. Violations can result in Notices of Penalty (NOPs) and financial penalties. Before CLE registration, there was no mechanism for a regional entity to audit a data center's compliance with reliability obligations, because the data center had none. After registration, a CLE that fails to respond to a curtailment signal under defined emergency conditions will be subject to enforcement action, just as a generation owner that fails to maintain required reactive support is subject to enforcement.

Fourth, the entity must participate in reliability planning processes. This includes providing data for annual Load and Resource Adequacy assessments, transmission planning studies, and regional transmission plans. A CLE owning 500 MW of connected load is a material input to a regional transmission plan. Before the CLE registry, that load might appear as a planning assumption but the entity owning it had no seat at the planning table and no obligation to validate those assumptions.

Fifth, and most consequentially for grid operators, registration enables the system operator to count on the CLE as a controllable resource under defined conditions. This is the difference between a demand object and a reliability resource. Under the CLE framework, a large compute load that can interrupt a defined block of consumption for a defined period becomes a tool in the operator's toolkit, not just a threat to manage.

Preparing Your Utility for CLE Registration Obligations

If you are a transmission owner, transmission planner, or reliability coordinator whose footprint includes large compute loads, the December 31, 2026 commitment date is your planning horizon. There is a set of preparation tasks that should already be underway. If they are not, this lesson is your call to action.

The first task is inventory. You need to know which loads in your system will qualify as CLEs once NERC defines the threshold. The March 2026 filing does not specify a megawatt threshold, but the industry expectation, based on comments filed in the FERC proceeding and NERC's alert language, is that the threshold will align with or reference the FERC large-load threshold of 20 MW. Some stakeholders have argued for a higher threshold, in the range of 50 MW to 100 MW, to limit the compliance burden on smaller facilities. Regardless of where the final threshold lands, the responsible preparation is to identify every interconnected facility that draws more than 20 MW and begin the data-gathering process that CLE registration will require.

The second task is data pipeline readiness. CLE registration will require metering and telemetry data that may exceed what your current billing and SCADA systems capture. Specifically, you will need interval demand data at a resolution of five minutes or better, load forecast data that the CLE submits on a forward-looking basis, and records of any curtailments or demand response events. If your metering infrastructure at large commercial and industrial substations is not capable of five-minute interval reporting with data retention adequate for compliance audits, you have a capital and configuration project to scope now.

The third task is tariff and agreement review. Your large-load interconnection agreements, your large-load tariff rate schedules, and any special contracts with data center customers will need to be reviewed for alignment with CLE obligations. Specifically, you need to understand what curtailment rights you currently have over large loads, whether those rights are sufficient to serve as the basis for CLE emergency curtailment compliance, and whether any existing agreement terms conflict with or supersede CLE obligations. This is work for your regulatory and legal teams, and it takes time.

The fourth task is planning model updates. If your transmission planning models and your load-flow studies treat data center loads as fixed injections with no flexibility, you will need to update those models to represent CLE loads as potentially controllable resources with defined operating characteristics. This changes the N-1 contingency analysis for substations serving CLEs, and it may change the transmission reinforcement projects that your planning process has in queue.

The Flexibility Asset Inversion: From Threat to Resource

The most strategically important implication of the CLE registry is what it enables on the operations side. Power system operators have long understood that a large, sudden load increase is a threat to frequency stability and voltage control. A 300 MW data center that goes from 50 MW to full draw in two minutes can cause a frequency excursion on a weak transmission system. That is why interconnection studies for large compute facilities spend significant effort on motor starting, voltage stability, and fault current calculations. The facility is modeled as a thing that happens to the grid.

The CLE framework inverts this relationship. If a 300 MW compute facility is registered, has provided its operating characteristics, and has agreed to curtailment protocols, it can be modeled as a reserve resource. Under grid emergency conditions, a 300 MW controllable load is equivalent, from a frequency response perspective, to 300 MW of spinning reserve. It responds faster than most combustion turbines. It can be ramped more precisely than most hydro units. And unlike generation reserves, which carry CAPEX and OPEX that must be recovered in rates, a CLE's flexibility may be available for a contracted payment or a tariff credit, potentially at lower cost.

This is not a hypothetical. The pattern is already visible in large-load demand response programs that several RTOs have developed. PJM's Emergency Load Response Program and ERCOT's Emergency Response Service are earlier, smaller versions of what the CLE framework generalizes. What the CLE registry adds is a standardized, reliability-obligated, auditable version of that relationship that applies system-wide and carries enforcement consequences if the CLE fails to perform when called.

For a utility's resource adequacy team, this means that the arrival of a 500 MW data center campus in your service territory, which used to be purely a transmission planning and interconnection problem, now has a potential upside: 500 MW of interruptible load that improves your resource adequacy position. For a control room operator, it means that the curtailment signal to the CLE is a tool in the operating procedure, not a negotiation. For a compliance lead, it means that the data center is now inside the compliance perimeter, with obligations that the operator can verify and enforce.

A registered large load is not just a demand to serve. It is a reliability resource with obligations on both sides of the interconnection agreement.

Worked Example: Onboarding a 500 MW CLE

Consider a transmission owner in a Mid-Atlantic state. In 2024, a hyperscale AI company signed a power purchase agreement and interconnection agreement for a 500 MW campus. At that time, the facility was treated as a large load customer: interconnection studies were completed, transmission reinforcements were agreed upon, and the facility began construction. The compliance relationship was a distribution or transmission service agreement with a standard large-load tariff rate.

Fast-forward to late 2026. The CLE registry is live. The transmission owner must now determine whether this facility meets the CLE threshold and, if so, facilitate the registration process. Here is what the onboarding workflow looks like in practice.

Step 1: Threshold determination. The transmission owner reviews the facility's metered peak demand, confirmed at 487 MW at the most recent summer peak. This exceeds the expected CLE threshold. The transmission owner notifies NERC's Registration Group and initiates the CLE application.

Step 2: Data submission. The CLE applicant (the data center operating entity) submits its registered entity data form. This includes legal entity identification, the NERC ID of the Transmission Owner and Balancing Authority it connects to, metering data for the prior twelve months, a demand forecast for the forward twelve months broken out by month, a description of any load management capability (the ability to reduce consumption by a defined amount within a defined response time), and the contact information for the entity's compliance officer.

Step 3: Agreement alignment. The transmission owner reviews its interconnection agreement with the data center for curtailment rights. It finds that the existing agreement allows the utility to curtail the facility for up to four hours per event under defined system emergency conditions, but does not specify a response time. The CLE standard, as drafted, requires a response within ten minutes for a Level 1 curtailment and within two minutes for a Level 2 emergency curtailment. The transmission owner initiates an agreement amendment to specify these response requirements and to add the telemetry interface that will allow the control center to send and confirm curtailment signals.

Step 4: Telemetry integration. The control center's EMS (Energy Management System) is updated to display the CLE's real-time demand, its availability status for curtailment, and any active curtailment signal. The data center's building management system is configured to receive and acknowledge curtailment signals via a defined protocol. End-to-end testing is conducted in a planned maintenance window.

Step 5: Planning model update. The transmission planner updates the power flow model to represent the CLE as a controllable load block with defined curtailment capability. The N-1 contingency analysis for the substation serving the campus is re-run with and without the curtailment capability active. The results show that the curtailment capability eliminates a voltage violation that would otherwise require a $45 million reactive compensation project. That project is removed from the capital plan pending confirmation of CLE curtailment performance.

Step 6: Compliance activation. The CLE receives its NERC ID and becomes a registered entity. The Regional Entity schedules its first compliance audit for eighteen months after registration. The transmission owner updates its compliance monitoring program to include the CLE's performance data in its own evidence records, since the transmission owner has obligations related to the accuracy of its planning assumptions and the reliability of its interconnection agreements.

This worked example illustrates something important: CLE registration is not just the data center's compliance event. It is also a transmission owner event, a planning event, and an operations event. The entire institutional relationship between the utility and its large compute customer changes when registration occurs.

The Standards Development Process and What Comes Next

It is worth being precise about the timeline, because the energy industry has learned the hard way that "committed by December 31, 2026" does not mean "enforceable January 1, 2027." NERC's Standards Development Process (SDP) has multiple phases: drafting, industry comment, ballot, and FERC approval. FERC must accept or approve a NERC standard before it has enforcement authority. For a standard as significant as the CLE registry, the full SDP cycle from first draft to FERC approval typically takes eighteen to thirty-six months.

What the March 2026 filing committed to is the initiation of that process with a delivery date for a final standard proposal. The enforcement date will depend on when FERC acts. Based on FERC's recent pace on interconnection rulemakings, utilities should plan for CLE standards to be enforceable no earlier than mid-2027 and potentially not until 2028. This does not reduce the urgency of preparation. Compliance audits cover historical practice. If your utility has not been maintaining adequate metering data or has not reviewed its large-load agreements for curtailment provisions, a 2028 audit can reach back to 2026 practices and find deficiencies.

The SDP also means that the final CLE standard will look somewhat different from the March 2026 filing. Industry comments, NERC's Standards Drafting Team (SDT), and FERC's review will all shape the final requirements. The most actively contested questions in the comment period are likely to be: the applicable standard set (which of the one hundred-plus NERC standards apply to CLEs), the megawatt threshold, the curtailment protocol requirements, and the data reporting cadence. Utilities that participate in the comment process will have better visibility into where the standard is heading and more influence over outcomes that affect their operations.

For a NERC compliance lead at a utility, the right posture right now is: monitor the Standards Development Process through NERC's public project pages, identify a subject matter expert within your organization who will track the CLE SDT, document your current large-load inventory and metering capabilities, and begin the internal conversations about agreement amendments and telemetry upgrades before they become deadline-driven fire drills.

Key Takeaways

• The NERC Computational Load Entity (CLE) registry, committed in a March 2026 FERC filing for delivery by December 31, 2026, creates a new registration category making large compute loads registered grid actors with enforceable reliability obligations, not merely demand objects.
• A May 2026 NERC Level 3 Alert formalized the reliability concern about large unregistered computational loads, signaling to registered entities that concrete action was expected, not advisory response.
• Registration carries five practical consequences: mandatory operational data sharing, applicable NERC standard compliance, audit and enforcement exposure, participation in reliability planning processes, and potential status as a controllable resource under emergency conditions.
• Utility preparation requires four parallel workstreams: inventory of threshold-meeting loads, data pipeline readiness (five-minute interval metering at minimum), tariff and agreement review for curtailment provisions, and transmission planning model updates to represent CLEs as controllable load blocks.
• The CLE framework inverts the large-load relationship from threat to resource: a registered 500 MW compute facility with curtailment obligations is equivalent to 500 MW of demand-side reserve, with potential capital deferral value of tens of millions of dollars in reactive and transmission reinforcement projects.
• NERC's Standards Development Process means the final CLE standard will be shaped by industry comment and FERC review; enforcement authority is unlikely before mid-2027 at earliest, but compliance audits can reach back to current practices, so preparation should not wait for the enforcement date.
• Utilities that proactively engage the CLE Standards Drafting Team will have better visibility and more influence over the final threshold, curtailment protocol, and reporting requirements than those who wait for the final rule.