U.S. DOE: Resource Adequacy Report: Evaluating the Reliability and Security of the United States Electric Grid. July 2025: Summary, Review, and Commentary

      In July 2025, the U.S Department of Energy released its ‘Resource Adequacy Report: Evaluating the Reliability and Security of the United States Electric Grid.’ In the executive summary, it is noted that firm, dispatchable generation was not being added fast enough to replace retiring coal plants and older natural gas units. They state that an intervention is necessary to meet growing demand from AI data centers, manufacturing, and re-industrialization. The latter two have yet to manifest. The DOE’s report relies heavily on the North American Electric Reliability Corporation (NERC) Interregional Transfer Capability Study (ITCS). The focus of the methodology of the report is on power demand, assumed load growth, assumed supply in light of retirements, magnitude of outages, and duration of outages. 

     Outage magnitudes utilize a metric known as Normalized Unserved Energy (NUSE), which is not standard in the U.S. but is being used more and more as an indicator. The duration of outages is measured in Loss of Load Hours (LOLH).

     The DOE has focused on an upgraded reliability standard, as described below:

“DOE’s methodology recognizes that the traditional 1-in-10 loss of load expectation (LOLE) criterion is insufficient for a complete assessment of resource adequacy and risk profile. This antiquated criterion is not calculated uniformly and fails to adequately account for crucial factors such as the duration and magnitude of potential outages. To provide a comprehensive understanding of system reliability and, specifically, to complement current resource adequacy standards while informing the creation of new criteria, the methodology uses the following reliability standard:

• Duration of Outages: No more than 2.4 hours of lost load in an individual year. This translates into one day of lost load in ten years to meet the 1-in-10 criteria.

• Magnitude of Outages: No more than an NUSE of 0.002%. This means that the total amount of energy that cannot be supplied to customers is 0.002% of the total energy demanded in a given year.”

     DOE ran reliability simulations based on these metrics. Below are LOLH and NUSE maps and some key findings of the plant closures case vs. the no plant closures case.

     The report notes:

“By 2030, without considering any generation retirements, an additional 12.5 GW of generating capacity is needed across PJM, SPP, and SERC to reduce shortfalls.”

     DOE’s methodology considered the availability of each generation source, outages, derates, load growth, AI growth, import/export capability, storage, demand response, retirements, additions, and idealized, or perfect capacity required.

     The report goes on to assess concerns in each region, comparing planned retirements vs. no retirements. The reality will likely be somewhere in between these scenarios. Data analysis for the PJM region is shown below.

     Below is a comparison of the existing generation stack with the different scenarios, including planned retirements and no retirements.  

     They do seem to make a good case that some plant retirements should and will be delayed, and these are mainly coal plants. I also know of an oil-burning plant in Baltimore whose retirement was delayed. I think a lot of the concern is with the availability of baseload or otherwise dispatchable power. Generally, each region has winter and summer reliability plans, resource adequacy protocols, and redundancies. However, as we found out during 2021’s Winter Storm Uri, sometimes the problem is too big for resource adequacy reserves to handle.  

     I also believe that amendments need to be made in cases where AI data center load growth is projected to be high enough to trigger power cost rises. In those cases, consumers should be protected from having to contribute to this growth, especially residential customers, but all consumers as well. The same could be said to be true for intermittent renewables integration. Those costs should be paid by developers and utilities, not other customers.

     There is certainly some concern about future resource adequacy, and there is also uncertainty about how much load growth AI data centers will ultimately need. There are, of course, other load growth drivers such as electrification and EVs, manufacturing and re-industrialization, and cryptocurrency mining.

     In August 2025, the DOE, citing persistent grid stress, extended three critical reliability orders, part of the emergency orders derived from Section 202(c) of the Federal Power Act. These three extensions were issued for plants in Pennsylvania, Michigan, and Puerto Rico. DOE also argued that closing these plants could raise power costs to customers served by those plants. Environmental groups and some public officials have opposed the retirement delays.

 

References:

 

Resource Adequacy Report: Evaluating the Reliability and Security of the United States Electric Grid. U.S. Dept. of Energy. July 2025. DOE Final EO Report (FINAL JULY 7).pdf

Federal Grid Interventions Enter a Second Phase as DOE Extends Emergency Orders. Sonal Patel. Power Magazine. August 28, 2025. Federal Grid Interventions Enter a Second Phase as DOE Extends Emergency Orders

US Plans to Use Emergency Powers to Stop More Coal Closures. Ari Natter. Bloomberg. September 24, 2025. US Plans to Use Emergency Powers to Stop More Coal Closures

Communities unite to oppose coal plant policy keeping expensive operations running: 'Unjust and unreasonable'. Tina Deines. The Cool Down. October 30, 2025. Communities unite to oppose coal plant policy keeping expensive operations running: 'Unjust and unreasonable'