October is not supposed to be emergency season. The planning year is built around summer peaks and winter storms, with the autumn shoulder months reserved for the maintenance that keeps the fleet ready for both. That is why the week of October 13, 2026, deserves more attention than a routine tight week usually gets: PJM Interconnection, the operator of the largest competitive wholesale market in North America, was operating under a federal emergency order, with demand response dispatched and maintenance recalled, because unusually warm fall weather collided with the maintenance season itself. The lights stayed on; the margin, and the tools required to hold it, is the story.
It is also only one third of the story. The same year that pushed PJM into emergency procedures in October had already pushed the Southwest Power Pool to its highest emergency alert in July and had tested the Electric Reliability Council of Texas through a record summer without a statewide conservation appeal. Read together, the three episodes form a kind of natural experiment. Three grids, facing versions of the same pressures — heat, load growth, planned outages, and variable output — produced three different outcomes. What separated them was not luck. It was the shape of each system’s vulnerability, and the shape of its resilience.
PJM: when the calendar becomes the contingency
PJM’s October squeeze began as a scheduling problem and became an adequacy problem. In its application to the Department of Energy (DOE) on October 11, PJM forecast demand of approximately 111,500 MW or higher on October 13, with roughly 61,591 MW of generation outages, largely planned work during the autumn outage season. Conditions on October 14 were projected to come within roughly 900 MW of the reserve margin. Temperatures ran more than 25 degrees above normal in parts of the region. The most acute pressure, PJM said, was in the Baltimore Gas and Electric, Pepco, and Dominion zones — the load pocket where mid-Atlantic demand, transmission constraints, and firm export obligations converged.
PJM layered its standard emergency procedures over the week. It issued a Maximum Generation Alert and a Load Management Alert for October 13, extended them through October 14, and declared an Energy Emergency Alert Level 1 (EEA 1) on October 13. It dispatched pre-emergency demand response on October 13 and again on October 14, covering the afternoon peak and the evening hours after solar output falls to zero, and it recalled all active generation maintenance outages effective noon on October 14. None of those steps required action by ordinary customers, and nothing in PJM’s public postings indicates the system reached EEA 3, the level at which firm load interruption is imminent or in progress. The emergency playbook worked in the sense that it was opened early and used in sequence.
What made the week different was the federal order behind it. On October 12, DOE issued Emergency Order No. 202-26-50 under Section 202(c) of the Federal Power Act, effective October 13 through October 16, directing PJM to dispatch specified generating units as needed for reliability and authorizing PJM to direct backup generation resources to operate as a last resort before or during an EEA 3. PJM’s vulnerability, in other words, was not a shortage of tools. It was a shortage of slack. When roughly 61,000 MW is offline for maintenance and construction, and key transmission outages are layered on top, a warm week stops being a shoulder-season inconvenience and starts consuming the reserve margin itself. PJM has said it is now experiencing capacity issues during milder weather, and expects those instances to increase as demand growth outpaces new supply. October was that warning in operational form.
The deeper lesson of PJM’s week is about the maintenance window. Shoulder seasons exist because the system needs them: plants must be inspected, repaired, and upgraded, and transmission must be built, and there are only so many low-load weeks in which to do it. That logic holds only while shoulder-season load stays low. As load grows — data centers, electrification, and hotter autumns among the drivers — and as the maintenance backlog grows with it, the window narrows from both sides. A system that cannot find a quiet week to repair itself will eventually have a loud week in which it cannot repair itself, either. PJM’s October did not prove the fleet is inadequate in the planning sense. It proved the calendar no longer provides the breathing room the plan assumes.
SPP: the alert that stopped one step short
If PJM’s test was about the calendar, SPP’s was about geography and market depth. On July 20, 2026, SPP’s western balancing authority escalated from EEA 1 at 4:38 p.m. Central Time to EEA 3 by 5:07 p.m., as a significant drop in solar generation coincided with the region’s highest peak demand of the summer season. EEA 3 is the highest level in the framework: operating reserves below required minimums, reserve-sharing assistance underway, and firm load interruption foreseen or implemented. SPP did not have to direct firm service interruptions that evening; imports eventually helped close the gap, and the authority stepped back down through EEA 2 and EEA 1 to Conservative Operations by 10 p.m. the same night. A second emergency followed on July 24, when a 769 MW reduction in nonfirm energy deliveries during the 8 p.m. hour prompted an EEA 2 declaration, again resolved without directed controlled interruptions.
The federal response followed the same pattern as PJM’s. DOE issued Order No. 202-26-36 on July 20, effective July 20 through July 21, authorizing SPP to dispatch specified generation units and to direct backup generation as a last resort before or during an EEA 3, and followed with Order No. 202-26-37 on July 26, in effect through August 3.
The sequence matters because it shows the limit of emergency authority as a planning substitute: the orders widened what SPP could dispatch in the moment, but they did not create the capacity that was missing from the market. SPP later told the Federal Energy Regulatory Commission that its western footprint has no resource adequacy requirements until June 1, 2027, and that this gap had significantly contributed to its advisories and emergency alerts — with about 9.2 GW of nameplate capacity in the relevant pool but only about 4.7 GW made available from April through August, against roughly 6.5 GW expected.
SPP’s vulnerability, then, was structural in a way PJM’s was not. PJM had a fleet largely present but temporarily offline for maintenance; SPP West had a market that, on the tightest evenings, did not bring enough energy to the table in the first place. That distinction explains why SPP’s alerts escalated faster and further: when the missing megawatts are not merely under repair but absent from offers, and when a drop in solar output arrives at the same hour as the peak, the operator’s remaining levers are imports, emergency purchases, and the alert ladder itself. The system held — no firm load was shed on July 20 or July 24 — but it held at the top of the ladder, with the next step being the one every operator hopes never to take.
There is a second, quieter lesson in SPP’s summer. The July 20 event turned on simultaneity: solar falling, demand peaking, and nonfirm deliveries shrinking in the same hours. That is the correlated-failure pattern this desk’s extreme-weather-planning explainer warned about — not one plant tripping, but many conditions deteriorating together. SPP’s operators responded quickly, and the de-escalation the same evening is genuinely to their credit. But a system that reaches EEA 3 in under half an hour has very little time for credit to matter: speed of response is not a substitute for depth of supply.
ERCOT: the summer that did not become an emergency
ERCOT provides the counterpoint, and it is worth stating precisely what it did and did not prove. Through a summer of record-breaking temperatures and a new peak demand record of 91.3 GW, ERCOT issued no statewide conservation appeal and declared no grid-wide emergency — a sharp contrast with 2023, when six such appeals were issued during extreme summer heat. Solar, wind, and battery storage carried much of the difference. According to an Environmental Defense Fund analysis of the summer, ERCOT added 95 percent of its new capacity from solar, wind, and battery storage from 2020 to 2025, for a total of 95 GW of such capacity, and for 94 percent of summer days the grid was powered by majority clean energy for at least one point in the day. Monthly solar often exceeded 18 percent of generation in peak summer months, with batteries discharging into the evening ramp as solar faded.
That is a real achievement, and it is also a bounded one. ERCOT’s summer success was a daytime and early-evening success: solar shaving the afternoon peak, batteries bridging the ramp, wind contributing overnight. It did not repeal the grid’s underlying constraints. ERCOT remains a largely isolated system with limited ties to its neighbors, so it cannot import its way out of a shortfall the way SPP did on July 20; its calm, hot evenings — when solar is gone, wind may be light, and thermal generation and storage must carry the load — remain its highest-stress hours.
And a summer without a conservation appeal is not the same as a winter-proof system: the February 2021 freeze remains the canonical example of correlated cold-weather failure, and no summer record erases it. What ERCOT proved in 2026 is narrower and still important: a fleet built heavily from solar, wind, and storage, dispatched flexibly, can absorb incremental load growth that would previously have triggered public appeals.
The contrast with PJM sharpens the point. ERCOT’s new capacity arrived in the same years PJM’s supply growth lagged its demand growth, and it arrived in forms — solar and four-hour-scale storage — matched to the shape of summer peaks. PJM’s October problem was a shoulder-season, transmission-constrained shape, where deliverability mattered as much as total megawatts. ERCOT’s lesson is that building the right shape of capacity, quickly, changes the emergency count. PJM’s lesson is that the shape of the emergency is itself changing, moving outside the season the fleet was built for.
What the three outcomes prove together
Put the three grids side by side and a pattern emerges that no single episode shows on its own. PJM had procedures and federal authority but too little slack in the maintenance season. SPP had neither a resource adequacy requirement in its western footprint nor enough energy offered on the tightest evenings, and reached the top of the alert ladder. ERCOT had a fast-growing fleet matched to its peak shape, and never needed the appeal at all. The common thread is not heat: heat was present in all three cases. The common thread is margin — how much of it each system held, in what form, and whether it could be delivered to the place and hour that needed it.
That reframes the planning question this desk raised in its extreme-weather explainer. The old planning assumption was that the future would resemble the past, and the new reality is that weather, load growth, and maintenance needs are all shifting at once. The 2026 episodes add a corollary: adequacy is no longer only a seasonal question. A system can pass its summer-peak test and fail its October test, as PJM nearly did; it can pass its energy test on most days and fail its evening-ramp test on one, as SPP nearly did; and it can pass its summer test while its winter test remains ungraded, as ERCOT’s summer leaves open. Planners who still size the system to a single peak, in a single season, are measuring the wrong thing.
There is also a governance lesson, and it cuts against easy narratives in both directions. Emergency orders did not fail in 2026 — in PJM and SPP they widened the operator’s options at the moment of stress, and in neither case did the system proceed to firm load shedding.
But emergency orders also did not solve anything durable: they are temporary, they expire, and repeated orders for the same region start to look less like emergency response and more like an operating mode. A grid that needs federal emergency authority in July and again in October is not a grid in crisis. It is a grid whose planning margins, market rules, and maintenance calendar have drifted out of alignment with its load — and emergency authority cannot realign them. That work belongs to capacity accreditation, transmission build, resource adequacy requirements, and maintenance scheduling in a system with fewer and fewer quiet weeks.
The honest bottom line is the one the extreme-weather explainer reached, now with 2026 evidence attached: no planner can weather-proof the grid against everything, and the question is which risks to buy down and at what price. PJM must buy down its shoulder-season exposure — more deliverable capacity in its constrained pockets, or a maintenance calendar that survives a warm October. SPP West must buy down its market-depth exposure before its June 2027 adequacy requirements arrive, or accept that its alert ladder will keep doing the work its market should do. ERCOT must buy down its evening and winter exposures while its summer success is still fresh. Three grids, three different purchases. The stress test did not crown a winner. It priced the homework.