With over a week to digest it, grid planning experts in interviews said the U.S. Department of Energy’s recent report on grid reliability overestimates demand growth and underestimates likely supply additions with the goal of preventing power plant retirements. (See DOE Reliability Report Argues Changes Required to Avoid Outages Past 2030.)

The report includes 50 GW of data centers, which likely exceeds the supply of chips that would be needed to build them, Grid Strategies Vice President Michael Goggin said. (See Doubt Cast from Different Angle on Data Center Load Demand.)

“They also assume 51 GW of non-data center load growth, and that’s pretty high, much higher than other projections that are out there, and particularly after the recent bill gutted incentives for electrification as well as for cleantech manufacturing in this country,” Goggin said.

A DOE spokesperson said its load growth assumptions are based on NERC’s Interregional Transfer Capability Study, with the addition of 50 GW of data center load picked as a midpoint from 2024 studies by the Electric Power Research Institute and the Lawrence Berkeley National Laboratory.

“Using a single planning midpoint addresses concerns of double counting and enables consistent load allocation across national transmission and resource adequacy models,” the spokesperson said.

The report’s prediction of 104 GW in generator retirements comes from NERC’s Long-Term Reliability Assessment released in December and the Energy Information Administration’s Annual Energy Outlook earlier this year. Both assumed that EPA regulations such as the greenhouse gas rules under Clean Air Act Section 111(d), which the Trump administration is actively working to overturn, will stay in place, Goggin said.

It also assumes additions of 20 GW of natural gas, 31 GW of four-hour batteries, 124 GW of new solar and 32 GW of new wind. It based them on NERC’s projections of “Tier 1” assets — those in development most likely to be completed. But Goggin and others said more capacity than that will be built by 2030.

“It also doesn’t appear to account for the contributions of renewables to providing output during peak demand periods,” Goggin said. “Wind and solar — solar more in the summer, wind more in the winter — provide dependable capacity value, just like other resources.”

Overall, it seems like the report assumes that markets and states’ integrated resource plans are not going to respond to load growth in the next five years beyond what is already in place, GridLab Executive Director Ric O’Connell said.

“It assumes NERC Tier 1 capacity additions, which is basically, as the report says, projects built in 2025 that are going to come online in the next two years,” O’Connell said. “And, so, it essentially assumes that nothing’s going to get built in 2027, 2028, 2029 and 2030, which is just not realistic.”

A DOE spokesperson said the report’s use of Tier 1 generation additions was grounded in reliability planning principles.

“DOE aimed to model a conservative yet realistic baseline. This approach is consistent with how NERC and planning authorities assess near-term reliability risks,” they said. “While we recognize that many additional resources are advancing through utility IRPs and interconnection queues, we also note that there is considerable risk and supply chain delays when it comes to dispatchable generation with lead times in many cases as far out as 2030.”

Even before the report came out, it was clear the administration was focused on keeping old fossil fuel power plants online.

“They’re retiring for a reason,” O’Connell said. “They’re uneconomic. They’re old. And instead of thinking about building new, they’re thinking that the only way to save the grid is to keep old stuff online. And I just think that’s not really what most utilities and markets are thinking about.”

PJM, MISO and SPP all have enacted rule changes to speed up new capacity additions, while utilities outside of the markets are actively addressing load growth through state regulations, he added.

“I think that’s one of the things the report also misses … [the] self-correcting, inherent nature of both power markets as well as the regulatory constructs … around resource adequacy,” Goggin said.

Higher prices from narrowing reserve margins are helping to bring new resources online and keep existing power plants that would have otherwise retired, he added. Vertically integrated utilities have their own mechanism addressing the same issue with state oversight and IRPs.

“State commissioners are certainly aware of the load growth and are making plans accordingly,” Goggin said.

O’Connell said that ultimately, the answer to DOE’s concerns is to get new resources online.

“We’ve got terawatts of capacity sitting in interconnection queues that haven’t been coming online,” he said. “Let’s get that capacity online. Let’s focus on streamlining the interconnection process, building new transmission, getting permitting reform right — clear the roadblocks for getting new capacity online. I feel like that the administration’s answer — ‘Let’s just keep these 60-year-old plants online’ — is just not the right answer.”

What Will DOE Do with its Report?

“It was fairly underwhelming,” Advanced Energy United’s Mike Haugh said. “It didn’t give any recommendations. It felt like the whole idea of this is a setup to basically issue more of the [Federal Power Act Section] 202(c) emergency filings.”

DOE recently used its power under the section to order the Campbell coal plant in Michigan and the Eddystone plant in Pennsylvania, which can burn natural gas or oil, to remain online. The Campbell order is being appealed. (See Order to Keep Campbell Plant Running Challenged at DOE and FERC.)

The report includes different scenarios, but the one with the highest reliability has no power plants closing for the next five years, which is why Haugh thinks DOE could use it to issue more such orders. That could happen with the Campbell and Eddystone plants because 202(c) orders are limited to 90 days.

Demand growth is contributing to tighter reserve margins around the country, which in organized markets are leading to higher prices that send the signal that more power plants are needed, but it is running into the fact that new plants take time to build.

“There’s a little bit of a lag,” Haugh said. “But it should incentivize some of these units to stay open a little bit longer. The problem is, some of these are so inefficient and they’re getting the capacity prices. … They’re not actually running the plant very often.”

So, while the natural market reaction will be to keep some power plants running longer than they otherwise would, others are too old and inefficient to bring in enough energy market revenue to stay open, and it will make economic sense to shut them down even with higher prices, he added.

The solution to the issue is clearing out the interconnection queues, Haugh said, which FERC and the industry already were working on before the new wave of demand growth came to dominate planning efforts. But that still can take up to five years, which is a snag in the process.

“You have projects that are ready to put steel on the ground,” Haugh said. “And you can get these combined advanced resources that can be built a lot faster than a gas plant.”

The industry already has regulatory mechanisms in place that have been working, and continue to work, to reliably meet the growing levels of demand, said Ari Peskoe, director of Harvard Law School’s Electricity Law Initiative.

“DOE has never played this role before, and it doesn’t need to try to play this role now, as sort of a master centralized planner,” Peskoe said. “It was sort of ironic from an apolitical faction that has historically kind of respected states’ rights on some of these issues.”

Peskoe noted that the report has a major disclaimer under the “Acknowledgements” section saying its analysis “could benefit greatly from the in-depth engineering assessments which occur at the regional and utility level,” where grid planners have access to better data.

DOE’s spokesperson explained that point further, saying: “The intent of the report is to complement, not override, the more granular, region-specific planning processes that incorporate a broader range of resources.”

“The bottom line is that the DOE team that wrote this paper acknowledges that its usefulness is very limited, and it should not supplant what happens at the regional level,” Peskoe said. “Because the utilities, RTOs, states and other entities involved in those decisions have better, more detailed information. So, I think that’s the most important takeaway from this paper.”

DOE’s main tool for addressing resource adequacy is Section 202(c), but its impact is limited to just 90 days and specific plants. The department also could try to get FERC to make some rule changes to stem retirements as it did in President Donald Trump’s first term, but that is just speculation, Peskoe said. And its main ability is to analyze the issues, which contributes to understanding the problem and developing solutions.

“If you look at the Biden administration, there was a lot of focus on transmission, and DOE put out a few reports about the country’s transmission needs, but they put those reports out after years of work, detailed consultation with industry and affected parties, and they were carefully done reports, whether you agree with them or not,” Peskoe said. “This was done in 90 days.”

Americans for a Clean Energy Grid Director Christina Hayes praised DOE for taking on the issue of load growth, which has dominated industry discussions for the past 18 months in part because of the uncertainty about how much of potential data center load will materialize.

“Generally, the way that this paper looks at the big challenges ahead of us is positive,” Hayes said. “What concerns me is that it tends to look backwards to the solutions. So, it’s thinking about it in terms of plants that are on the system, rather than how to plan going forward.”

That new planning will involve new generation coming onto the system, but it also will require more transmission to move power around a bigger power system. Winning the “AI race” is a bipartisan goal, and Hayes noted the U.S.’ competitors are investing in their grids.

“I think there was a statistic to something like in 2022, China invested $168 billion in their grid,” Hayes said. “The United States invested $22 billion in its transmission system. So, just on an apples-to-apples investment in the wires needed to support all of this new load and all of the new generation, we are far behind.”

Infrastructure investment is starting to ramp up, Congress could take another crack at permitting legislation in Trump’s term after the Manchin-Barrasso bill failed to advance last year, and some of the regions are moving forward with more transmission investment.

“We’re seeing it on the ground, with 765-kV lines being proposed in Texas to support the oil and gas industry and their needs for power,” Hayes said. “SPP, PJM and MISO are all looking at 765-kV lines to help support their greater electrification needs as well. So, we’re seeing the region start to move on it, not because it’s a partisan idea, but because it’s a good idea.”

Lines at 765 kV have rarely been used in the U.S., but they can help move more power and can avoid building out multiple lines at lower voltages. Another option for getting more electrons around is to use advanced conductors at lower voltages, Hayes said.