The North American Electric Reliability Corporation (NERC) has issued a Level 3 Alert, the highest possible severity level and one that is rarely used by NERC, requiring the immediate attention of many NERC registered entities. While the Alert is not itself a new Reliability Standard, it directs certain registered entities to assess and report on proposed changes to modeling, studies, protection, and communications practices for large computational loads.
The Alert is another incremental step in NERC’s broader effort to address the reliability implications of large loads, particularly data centers, artificial intelligence training facilities, cryptocurrency mining operations, and other computational loads whose size and operational characteristics may create reliability risks not fully captured in traditional planning practices.
The Alert arrives amid increased attention by FERC and NERC to the pace and scale of data center development. NERC is currently considering multiple long-term options to address grid risks associated with rapidly growing computational loads, including the possibility that certain large-load customers may ultimately become subject to new registration criteria and Reliability Standards.
In the interim, with the Alert NERC is taking immediate action to address reliability risks related to large computational loads by directing responses from entities that engage with those loads and are already registered with NERC. Those registered entities were required to acknowledge the Alert by May 11, 2026 and must report on the status of their activities by August 3, 2026.
Background
Computational loads are increasingly relevant to reliability planning as they can differ materially from traditional industrial load.
Large data centers and similar facilities may include substantial power-electronic equipment, uninterruptible power supply systems, sophisticated controls, cooling loads with power electronic drives, on-site generation, battery storage, and operating modes that allow the facility to rapidly reduce load, transfer to backup resources, or reconnect to the grid. These characteristics can affect voltage stability, frequency response, protection coordination, contingency analysis, and operator situational awareness.
Since 2022, NERC has analyzed numerous load loss events in the Eastern Interconnection and Texas Interconnection resulting from the unexpected disconnection of computational loads. Those losses were not predicted through current planning studies. In September 2025, NERC issued a Level 2 Alert addressing the risks associated with large loads broadly and recommending practices for large-load interconnection, study, commissioning, and operations.
Based on responses to its prior Level 2 Alert, NERC found that computational loads such as data centers could increase exponentially over the next four years and that significant risks to the Bulk Power System needed to be addressed through immediate industry action.
The Alert also follows several incidents in 2024 and 2025 in which more than a gigawatt of data center computational load reportedly disconnected from the grid at once. These customer-initiated large-load reductions occurred in seconds, leaving limited time for real-time operator response and raising reliability concerns.
The concern is not limited to load loss alone: similar reliability issues may arise when large computational loads come back online, transfer to or from backup generation, or change operating modes in response to voltage or frequency disturbances.
Applicable Entities and Deadlines
The Alert was initially distributed to entities registered with NERC for the functions of Transmission Planners, Planning Coordinators, Transmission Owners, Balancing Authorities, Reliability Coordinators, and Transmission Operators.
The Alert contains seven Essential Actions that NERC determined to be essential for certain registered entities to take to ensure the reliable operation of the Bulk Power System.
The Essential Actions all relate to modifying processes and procedures to address the risks to reliability posed by the rapid integration of large computational loads to the electric grid. However, the Essential Actions are not the same as Reliability Standards and are not mandatory obligations. Registered entities are not subject to penalties for failing to implement the Essential Actions.
All recipients of the Alert were required to provide initial acknowledgement of receipt via the NERC Alert System by May 11, 2026. These recipients must now provide responses to 33 questions related to the Essential Actions by August 3, 2026.
The Seven Essential Actions
Together, the seven Essential Actions are intended to improve visibility into computational load behavior and ensure that those characteristics are reflected in interconnection, planning, commissioning, protection, monitoring, and operating practices:
- Modeling data, settings, and parameters. Transmission Planners and Planning Coordinators should develop a detailed list of modeling data, settings, and parameters needed from computational loads and distribute that list to Transmission Owners in their footprint. Transmission Owners should reflect those requirements in their facility interconnection requirements. The Alert identifies information such as electrical size, power factor, dynamic characteristics, ramp rates, protective-device settings, IT versus non-IT load composition, on-site generation, and facility use.
- System studies for computational loads. Transmission Planners and Planning Coordinators should study their systems with additional consideration for computational loads. These studies should evaluate operating-envelope limitations before voltage or frequency instability, identify areas vulnerable to instability, develop mitigation measures, and update contingency files where aggregate loss or reduction of computational load could cause planning criteria violations.
- Qualified change triggers. Planning Coordinators should revise their definition of “qualified change” to account for computational loads. At a minimum, the definition should capture growth of the computational load, changes to settings or configurations of electrical supply or end-use equipment, and repurposing of a facility for a significantly different application.
- Commissioning processes. Transmission Owners should establish a commissioning process for computational loads. The process should include evaluation of the as-built model, model verification and validation, coordination with affected nearby systems, load and no-load testing where feasible, testing of relevant control modes, commissioning checklists, and post-energization verification of SCADA data, diagrams, ratings, and responsibility boundaries.
- Corrective actions and ride-through. Transmission Planners and Planning Coordinators should study and implement system-side corrective actions with Transmission Owners to avoid nonconsequential loss of firm computational load from normally cleared non-bus faults. This includes obtaining facility-level relaying and protection settings and evaluating how computational load protection and control systems respond to normally cleared faults.
- Dynamic fault recording. Transmission Owners should install or use dynamic fault recording devices to capture computational load facility performance during system disturbances. This may be done in coordination with the computational load owner, including by installing devices or obtaining access to existing devices, and the resulting data should be available for event analysis and root-cause analysis.
- Direct communications with computational loads. Transmission Operators, Reliability Coordinators, and Balancing Authorities should establish interpersonal communication capabilities with computational loads to improve situational awareness and support joint operating procedures during planned and emergency conditions. The Alert contemplates that these entities may issue instructions, orders, or similar information through voice, SCADA, or other communication platforms to prevent BES Emergencies.
Looking Ahead
The Alert does not impose new mandatory obligations to NERC registered entities or directly impact computation load customers, however, it portends the broader reliability initiative by FERC to make changes to the Reliability Standards, NERC registration requirements, and other rules and regulations to address reliability risks associated with large computational loads to the Bulk Power System.
While binding action on registered entities or direct impact, if any, on computation loads is a longer-term goal, the Alert serves as a preview to the concerns NERC has identified and some of the strategies it sees are necessary to maintain reliability.
For registered entities in scope of the Alert, there are near-term notification and response obligations. Registered entities should also consider whether existing modeling requirements, interconnection processes, study procedures, and other processes adequately address computational loads.
These considerations will vary by registered entity and should also reflect the geographic location of the registered entity to the extent local jurisdictions may also impose additional regulations related to computational loads.
For example, in ERCOT, that evaluation should also account for Texas Senate Bill 6 and the Public Utility Commission of Texas’s developing large-load interconnection framework, particularly where facilities have behind-the-meter generation, backup generation, or operational curtailment obligations that may affect planning models, emergency procedures, and responses to the Alert.
More broadly, the Alert reinforces the shift in NERC’s consideration of large computational load from primarily issues limited to load-forecasting or interconnection-queue to being treated as dynamic, reliability-relevant facilities whose size, operating characteristics, protection behavior, and interactions with the grid may lead NERC to require more detailed planning and operational oversight.