North American Electric Reliability Corporation (NERC)

NERC’s computational load registration at 50 MW and 100 kV creates demand for ride through, disturbance recording, and operational data capabilities that grid software and power electronics startups can build. Three Dominion zone events frame the market. About 1,500 MW of data center load tripped on July 10, 2024. About 500 MW stayed off the grid after the April 18, 2026 fault. About 3,800 MW disconnected on July 22, 2026, after a normally cleared fault on a 230 kV line (Sources 5, 9).

FERC ordered NERC on July 16, 2026, to file Reliability Standards and registration criteria for computational loads by December 31, 2026 (Source 1). The draft standards passed their final industry ballot on October 15, 2026, with affirmative votes from 77.93 to 95.45 percent against a 66.67 percent passing threshold (Source 6). The registration criteria cover sites of at least 50 MW connected at 100 kV or higher, raised from the 20 MW and 60 kV draft of April 2026 (Source 23).

The standards in the first filing require three capability groups. Owners must supply planning model data that includes tripping and ride through behavior. Owners and operators must follow operating instructions and keep communications open with grid operators. Owners must install sequence of events, fault, and dynamic disturbance recorders on the high side of the main transformer (Source 6). Minimum voltage and frequency ride through requirements are not in the first filing. They sit in a separate project that NERC’s Standards Committee approved on October 21, 2026 (Source 25).

Texas already applies a ride through curve to large electronic loads of 75 MW or more under NOGRR 282, effective August 1, 2026 (Source 16). The strongest product opportunities sit where the first filing creates a dated compliance duty and where the later ride through rules will need verified equipment. For investors and policymakers, the registered site count that NERC’s expedited data request will produce in December 2026 is the figure that sizes the market.

Background and History

How Did Computational Loads Become a Registered Reliability Risk?

Computational load became a registered reliability risk because a protective disconnection by one data center campus is harmless, while the same disconnection by dozens of campuses on one fault removes gigawatts of demand in seconds. NERC’s July 10, 2024 incident review documented the first event of that size. A failed lightning arrester on a 230 kV line in the Eastern Interconnection caused a permanent fault. Reclosing produced six faults in 82 seconds, each cleared in 42 to 66 milliseconds. Voltage in the affected area fell to between 0.25 and 0.40 per unit. About 1.5 GW of load, all of it data center type load, left the grid on customer side protection. None of it was disconnected by utility equipment (Source 22).

NERC found that about 1.26 GW dropped on the third voltage depression and stayed off for hours. The cause was a customer side counting scheme that trips after three disturbances in about one minute and holds the facility on backup power until someone reconnects it manually. Static UPS systems returned quickly once voltage recovered. Diesel rotary systems typically needed a manual transfer back (Source 22). Frequency rose to 60.047 Hz and took four minutes to settle, and voltage rose to 1.07 per unit while operators removed shunt capacitor banks (Source 22).

The industry response moved in steps. NERC issued a Level 2 alert on September 9, 2025, with five recommendations covering interconnection requirements, study processes, commissioning, operations, and demand forecasting (Source 3). NERC issued a Level 3 alert on May 4, 2026, its highest alert level, with seven essential actions and a response deadline of August 3, 2026. The Level 3 alert created no mandatory obligation and no penalty exposure (Sources 2, 17, 19). On April 1, 2026, NERC posted draft registry criteria of 20 MW at 60 kV and a Standard Authorization Request for computational load (Source 2).

FERC converted the voluntary track into a dated one on July 16, 2026. In Docket RD26 7, issued under section 215(d)(5) of the Federal Power Act, FERC directed NERC to file new or modified Reliability Standards and Rules of Procedure revisions with registry criteria by December 31, 2026, and a workplan for further standards by March 1, 2027 (Sources 1, 14, 15). The order is reported at 196 FERC ¶ 61,031 (Source 14). FERC’s separate large load interconnection proceeding addresses how a site connects. This report addresses what the connected site must ride through, report, and record.

Six days after the order, the July 22, 2026 event removed about 3,800 MW from PJM, the largest load transfer PJM had recorded, and PJM announced on August 11, 2026 that it was evaluating ride through requirements of its own (Sources 10, 20). The five proposed standards passed their first industry ballot, reported on September 22, 2026 (Source 24). NERC’s Standards Committee approved a further project on October 21, 2026, for ride through and active power recovery (Source 25). The load growth behind these events is large. The Lawrence Berkeley National Laboratory estimated 2023 data center electricity use at about 176 TWh, or 4.4 percent of U.S. electricity, and projected 325 to 580 TWh, or 6.7 to 12.0 percent, in 2028 (Source 11). NERC projects summer peak demand to rise by more than 224 GW over ten years (Source 5).

Detailed Findings

What Does the NERC Registration Rule Require of a 50 MW Computational Load Site?

The Established Position

A computational load site of at least 50 MW connected at 100 kV or higher must register as a Computational Load Owner, a Computational Load Operator, or both, and must meet standards CLO 001, CLO 002, and CLO 003 once FERC approves them. NERC defines a Computational Load Site as one or more buildings, structures, or installations at a single lo