In the third week of September 2026, the U.S. Department of Energy reached a milestone that would have been difficult to imagine only a few years ago. With directives issued between September 17 and 19, the agency’s emergency actions under Section 202(c) of the Federal Power Act pushed into the high forties for the year, affecting roughly seven aging power plants and eleven generating units once scheduled for retirement. A tool written for wartime and historically reserved for true short-term crises has become something closer to a recurring feature of grid operations, reissued on a rhythm that now tracks the calendar more than a specific emergency. The same week the newest orders landed, a federal appeals court delivered the sharpest rebuke yet to the legal theory behind them, setting up a reckoning over how the country manages the retirement of dispatchable capacity. For anyone tracking grid reliability, the coming months will test whether emergency authority can substitute for orderly resource planning, and at what cost.

The issue is no longer whether Section 202(c) can help in a true emergency. It can. The issue is whether repeated ninety-day orders are becoming a substitute for the harder work of resource adequacy, transmission planning, and state-level retirement decisions.

The September Escalation

The most recent wave of DOE 202(c) emergency orders reads like a map of the grid’s pressure points. On September 17, the department directed PJM Interconnection to keep specified units and backup resources available as the mid-Atlantic braced for stressed system conditions. A day later, it turned to the Carolinas, authorizing Duke Energy Carolinas to dispatch designated units through September 21 as late-season heat pushed demand higher across the Southeast. The same sequence reached into the Midwest, ordering Northern Indiana Public Service Company to keep Units 17 and 18 of the R.M. Schahfer Generating Station running, and directing CenterPoint Energy to hold Unit 2 of the F.B. Culley station in Warrick County available, both from September 20 through December 18. Each order authorized operation “notwithstanding” permit limitations and framed the intervention as a reliability backstop against retiring capacity and rising load.

What makes these orders remarkable is not any single directive but their accumulation. The Indiana orders, for example, are the fourth consecutive extensions for the same coal units, following orders issued in December 2025, March 2026, and June 2026. The plants were supposed to retire; instead, they have been carried forward in ninety-day increments, with each order arriving just before the last one lapses. The Duke Carolinas order, tied to a specific heat event, still resembles the kind of transitory emergency the statute was designed to address. The Indiana orders look different: long-term capacity planning conducted through emergency channels. That distinction is what has drawn the courts into the fight.

A Statute Stretched Beyond Its Origins

Section 202(c) of the Federal Power Act was enacted in 1935 and sharpened during World War II to give the federal government a lever for genuine, immediate emergencies: a sudden generation shortfall, a natural disaster, or a wartime disruption. For decades, it was invoked sparingly, often for days rather than months, and typically at the request of the grid operators or utilities responsible for running the system. The current pattern departs from that history on nearly every dimension. The orders are numerous rather than rare, measured in months rather than days, and in several cases issued without a request from the plant operator, the regional transmission organization, or the state regulator responsible for reliability planning. Sierra Club senior attorney Greg Wannier captured the shift bluntly earlier in the cycle, arguing that the administration had “twisted the use of this emergency authority beyond all recognition.”

The administration’s defenders frame the same facts as prudent stewardship in an era of surging demand. DOE Associate Deputy Secretary Alex Fitzsimmons has argued that “resource adequacy is worsening” and that the country “needs dispatchable generation” to manage the transition. That argument has real force in a broader environment where data center load growth, electrification, and a manufacturing revival are converging on a grid retiring firm capacity faster than it is replacing it. NERC’s long-term assessments have flagged mounting reliability risk, and DOE points to its emergency actions as preserving dispatchable megawatts during tight periods. The dispute, then, is not whether the grid is under stress. It is whether an emergency statute is the right instrument for managing a structural, multi-year mismatch between supply and demand.

The Courts Push Back

That question became binding law on September 11, 2026, when the U.S. Court of Appeals for the District of Columbia Circuit vacated DOE’s order delaying the retirement of Consumers Energy’s 1,420-megawatt J.H. Campbell coal plant in Michigan. The court’s reasoning struck at the heart of the strategy. An “emergency,” it held, requires immediate action to address a transitory crisis, not a standing response to long-term planning concerns. Reliability planning, the judges emphasized, belongs to the states, working with regional operators and utilities, and the department’s “sweeping conception” of its own powers could not be squared with the statute. DOE, the court noted, had built its case on “fragments” drawn from two documents and a MISO presentation, and its interpretation “invites frequent federal interventions that are unsupported by the statute and threaten the stability of the energy market.”

The timing could hardly have been more pointed. Nine days after the D.C. Circuit vacated the Campbell order as an overstep of federal authority, DOE issued fresh orders for Indiana coal units relying on the same Section 202(c) mechanism. The legal exposure is obvious. DOE has responded by reissuing orders before they expire and signaling that it may seek Supreme Court review. For now, the ruling clouds the footing of other plants operating under similar directives, most of them coal-fired, and strengthens the hand of state attorneys general, consumer advocates, and clean-energy groups challenging the orders. The December 18 expiration of the newest Indiana orders now looms as a real checkpoint for whether the coal-preservation strategy can survive judicial scrutiny.

The Bill Comes Due

Beneath the constitutional argument lies a more immediate question for utility customers: who pays to keep these plants alive? The costs are neither trivial nor fully transparent. An early Sierra Club analysis pegged the price of thirteen orders across six plants at more than $230 million, defining that figure as the marginal additional cost plant operators have asked consumers to absorb. As the orders multiplied through 2026, the estimate climbed past $547 million, with the group calculating net expenses at roughly $1.5 million per day. The Institute for Energy Economics and Financial Analysis separately tallied more than $300 million in added costs through mid-May. Individual plants tell the story in miniature: Consumers Energy reported roughly $401 million in costs tied to the Campbell plant through the end of March and has sought to recover $180 million from ratepayers, while TransAlta has pursued recovery for its Centralia facility in Washington along with additional repair expenses.

The recovery mechanism is what turns these figures into a live political issue. When a plant is ordered to run past its retirement date, the operator typically asks regulators to let it collect the extra expense through customer bills. A federal reliability judgment therefore lands on households and businesses that had no direct role in the intervention. That dynamic has produced unusual alliances. In Colorado, Tri-State Generation and Transmission and Platte River Power Authority filed a rehearing request arguing that an order forcing them to keep a coal generator running past its planned retirement infringed their rights, precisely because no one operating the system had asked for the intervention. The recurring complaint is not merely that the orders are expensive. It is that they socialize the cost of a contested federal policy while bypassing the state ratemaking processes designed to weigh exactly these tradeoffs.

The Reliability Question Underneath

If the plants were delivering clear reliability value, the expense would be easier to defend. The operational record complicates that case. Reporting on the units under emergency orders has found that several are producing significantly less electricity than they did before retirement was first scheduled, and by mid-year four of the eleven units under orders were not operating at all. During Winter Storm Fern in late January, two Schahfer units ran at more than 285 megawatts each on the coldest days, providing a concrete example of the firm output the orders are intended to preserve. Yet a subsequent FERC-NERC review of that same storm found no significant reliability failures across the region, and MISO’s capacity margins have generally exceeded targets, with anticipated additions projected to outpace demand growth for years. GridLab program director Nikhil Kumar has put the skeptics’ position plainly: “There is no emergency.” Even some affected operators have said the same in substance, with CenterPoint’s Indiana leadership acknowledging that the Culley unit “isn’t needed” for reliability.

The counterargument has merit. Reserve margins are backward-looking averages that can mask the specific hours when the system is genuinely tight, and data center load growth has repeatedly outrun official forecasts, leaving planners chasing a moving target. NERC officials have credited the emergency orders with helping maintain reliability during stressed intervals, and the prudent-risk case is straightforward: a modest insurance premium can be defensible when the alternative is firm load shedding. The harder conclusion is that both things can be true at once. The orders may buy real insurance during a handful of critical hours while also imposing costs and emissions that look excessive against the plants’ modest annual output, including millions of additional tons of carbon dioxide and substantial sulfur dioxide and nitrogen oxide emissions.

What It Means for Planners and Markets

For utilities, grid operators, and investors, the deeper damage may be to the predictability that resource planning requires. Retirement dates are not arbitrary; they anchor replacement generation, transmission upgrades, capital recovery, and workforce transitions. When a plant slated for closure is instead held open through a sequence of ninety-day emergency orders, every downstream decision wobbles. Capstone vice president Erin Melly has warned that the short order terms create uncertainty fundamentally misaligned with how grid operators forecast. An order that keeps a plant running for three months at a time is not a plan. It is a deferral, and deferral without a defined endpoint corrodes the reliability planning the orders claim to protect.

The market implications radiate outward from there. Capacity auctions, most visibly in PJM, already reflect a system straining to price scarcity, and the persistent presence of units operating outside normal market signals distorts the investment case for the new dispatchable generation everyone agrees the grid needs. If developers and financiers cannot trust that retirements will proceed on schedule, the risk premium on new projects rises at exactly the moment the country is trying to accelerate them. In that sense, the emergency-order era may be working against its own stated goal: propping up yesterday’s fleet in a way that makes tomorrow’s harder to build.

Conclusion

The September 2026 flurry of DOE 202(c) emergency orders, arriving just as a federal appeals court declared the underlying strategy an overreach, marks an inflection point rather than a resolution. The grid genuinely faces tightening conditions as data center load growth and electrification collide with an aging, retiring fleet, and no serious observer disputes that reliability must be protected. What is in dispute is whether an emergency statute, invoked on a ninety-day loop and paid for by ratepayers who never requested it, is the right way to protect it. The D.C. Circuit’s answer was no, at least as applied to long-term planning dressed up as crisis response. Whether that answer survives further appeal, and whether the December expirations bring orderly transition or another round of extensions, will tell the industry a great deal about how the United States intends to govern the most consequential grid buildout in a generation. The durable lesson is simple: emergency authority is a poor and expensive substitute for the patient work of planning, and the sooner the sector rebuilds a credible path for adding firm capacity, the sooner it can stop governing the grid ninety days at a time.