The Federal Energy Regulatory Commission (FERC) has expanded the federal reliability framework by directing the North American Electric Reliability Corporation (NERC) to develop mandatory reliability standards for computational loads, a category that includes artificial intelligence data centers, cryptocurrency mining facilities, and other large information technology operations. The Federal Energy Regulatory Commission (FERC) has ordered the North American Electric Reliability Corporation (NERC) to create mandatory reliability standards covering computational loads, a category that includes hyperscale AI data centers, cryptocurrency mining operations, and other high-density computing facilities.

The directive introduces a compliance framework that places unprecedented attention on how rapidly changing electricity demand from these facilities affects the stability of the Bulk-Power System. Rather than treating data centers solely as customers purchasing electricity, regulators are preparing to evaluate them as participants capable of influencing transmission reliability. The decision reflects growing concern that modern computing campuses can alter power consumption within seconds, creating operational conditions that traditional grid planning was never designed to accommodate. The order also establishes a definitive regulatory timeline that converts years of technical studies into enforceable federal obligations for infrastructure developers, utilities, and computational load operators.

Computational Loads Move Toward Direct Federal Accountability

FERC’s order, issued under Docket No. RD26-7-000 on July 16, requires NERC to submit new or revised mandatory reliability standards by December 31, 2026, alongside procedural changes that could register computational load entities directly under the reliability framework established through Section 215 of the Federal Power Act. Until now, mandatory reliability responsibilities primarily focused on generators, transmission operators, balancing authorities, and other traditional grid participants. The commission’s latest action creates the regulatory pathway for certain large electricity consumers to become directly accountable for maintaining system reliability if they satisfy future registry criteria. Those criteria will ultimately depend on physical and electrical characteristics rather than simply overall facility size, leaving NERC responsible for determining appropriate thresholds through its standards development process. Infrastructure developers planning large AI campuses must now consider regulatory compliance alongside transmission access, energy procurement, and facility construction schedules.

AI Demand Volatility Emerges as the Central Grid Challenge

The commission’s action reflects mounting evidence that next-generation computing facilities behave differently from conventional industrial electricity consumers. AI training clusters, cloud infrastructure, and cryptocurrency operations possess the ability to increase or reduce electrical demand almost instantaneously because workloads can shift rapidly across thousands of servers. Such behavior introduces operational challenges for transmission systems that depend upon predictable consumption patterns to maintain frequency and voltage stability. Moreover, regulators increasingly view these demand characteristics as system-wide reliability concerns rather than isolated customer operational decisions. The accelerating deployment of gigawatt-scale AI campuses across North America has intensified this discussion because utilities must now prepare for concentrated power demand capable of changing faster than many conventional balancing resources can respond.

Federal oversight therefore aims to ensure that AI expansion does not outpace the reliability safeguards required to support it. FERC Chairman Laura Swett explained the commission’s position during the agency’s open meeting, stating that “certain large loads like data centers and crypto mining operations that NERC calls computational load have the potential to change their demand almost instantly.” She added that “this rapid fluctuation causes voltage stability issues that threaten grid reliability.” Swett also emphasized that the commission’s decision removes uncertainty surrounding NERC’s implementation schedule, ensuring that the “critical work of winning the AI race does not threaten reliability in our country.”

NERC Research Forms the Foundation for Mandatory Standards

The commission’s directive builds upon more than two years of technical investigations conducted by NERC into the operational effects of rapidly expanding computational loads. Much of that research emerged from the “Interconnection of Large Loads to the Interstate Transmission System” proceeding, which examined how accelerated deployment of high-density computing facilities could influence grid operations across interconnected transmission networks. During those assessments, NERC documented operational scenarios in which customer-initiated load reductions occurred within seconds, leaving system operators with little opportunity to respond through conventional balancing mechanisms. Engineers concluded that these sudden demand changes could create conditions capable of threatening Bulk-Power System reliability under specific disturbance scenarios.

The findings shifted industry attention from long-term electricity demand forecasting toward the operational behavior of computational facilities during grid events. Those conclusions now underpin the mandatory standards development process ordered by FERC. NERC President and CEO Jim Robb previously referenced actual operating events demonstrating these risks, noting that “large data center loads have responded to, and amplified, grid instability.” Those observations, particularly from incidents in Virginia and Texas, reinforced concerns that computational loads can influence grid disturbances instead of merely responding to them.

Real Grid Events Strengthened the Regulatory Case

Federal regulators also relied upon documented operating events demonstrating how computational loads have interacted with transmission disturbances under real-world conditions. One investigation reviewed an event that occurred on July 10, 2024, within the Eastern Interconnection, where a transmission fault resulted in the near-simultaneous loss of approximately 1,500 megawatts of data center load. Another assessment examined cryptocurrency mining operations operating within ERCOT and concluded that these facilities could reduce electricity consumption by between 17% and 95% within milliseconds following transmission disturbances. NERC’s 2026 State of Reliability report further identified large computational loads as an emerging contributor to frequency and voltage instability across the North American grid. Consequently, regulators determined that voluntary guidance alone would no longer provide sufficient assurance as AI infrastructure continues expanding across multiple transmission regions. These documented operating experiences strengthened the technical justification for converting research recommendations into mandatory reliability obligations.

Phase I Establishes Standards While Phase II Expands the Framework

The commission has divided implementation into two distinct phases designed to accelerate initial protections while allowing additional technical work to continue. Phase I requires NERC to submit one or more new or revised reliability standards by December 31, 2026, covering essential operational actions for integrating computational loads into the Bulk-Power System. The same filing must include revisions to the NERC Glossary, where computational load is proposed to be defined as “load comprised of power demand from information technology equipment, such as servers, storage, and networking hardware.” NERC must also propose amendments to its Rules of Procedure, including new registry criteria that determine which computational load entities become subject to mandatory compliance requirements. Meanwhile, Phase II extends beyond the initial standards by requiring NERC to submit an informational work plan by March 1, 2027, describing the development of additional computational load reliability standards.

Compliance Planning Becomes a Strategic Requirement for AI Developers

The regulatory implications extend well beyond utilities because hyperscale developers, colocation providers, and digital infrastructure investors now face an evolving compliance environment alongside traditional construction and financing considerations. Future project planning for large computational load facilities will need to account for the mandatory reliability standards and registry criteria that NERC has been directed to develop under FERC’s order before those requirements take effect. Developers pursuing gigawatt-scale campuses may need to demonstrate not only adequate transmission access but also operational characteristics consistent with future reliability standards governing computational loads. Until the reliability standards are finalized, FERC has left the specific technical compliance obligations, registration thresholds, and implementation requirements to NERC’s standards development and governance processes.

The order establishes that certain computational load entities may become subject to mandatory reliability obligations once NERC develops the applicable registry criteria based on specific physical and electrical characteristics. The order marks the beginning of a formal standards-development process that will establish how qualifying computational load entities are incorporated into the mandatory federal reliability framework. Commissioner David Rosner summarized the broader policy objective behind the order, stating that the commission’s goal is to “get large loads online quickly, reliably, and with safeguards that protect regular consumers.” All five FERC commissioners voted in favor of the directive, reflecting unanimous support for establishing a federal reliability framework capable of accommodating the next generation of large-scale computational infrastructure while protecting overall grid stability.