Friday, October 16, 2026

What Needs Your Attention Today

Green means watching — we are tracking it and it looks normal. Amber means warning — it needs a closer look. Red means danger — time to act. The color never carries the meaning alone: every card says it in words.

Module 1 — The 60-Second Regime Matrix

Regime: emergency management works; emergency management is becoming routine. The claim: this week's real story is not that the grid nearly failed — it did not — but that keeping it comfortable now requires federal emergency powers in October. The mechanism: demand keeps growing (data centers are the fastest-growing load), while a large share of plants are offline each autumn for planned maintenance, so the margin that used to exist in the shoulder seasons — the mild weeks between summer and winter peaks — is gone. The evidence: PJM told the Energy Department it expected demand of about 111,500 MW on Oct 13, roughly 61,591 MW of generation unavailable, and a cushion of only about 900 MW above its required reserve on Oct 14 — on a system built for far more. PJM expected demand of 113,312 MW on Oct 14. Spot prices (last-minute power) at PJM's Western hub — its main wholesale pricing point for the western part of the region — hit $1,021 per megawatt-hour on Tuesday, and next-day prices jumped 28% to $295.83 — numbers you expect in a heat wave, printed in mid-October. What's priced in: Constellation closed Thursday at $275.00, still about 20% above its 52-week low of $228.63 — the market is paying for scarcity, but no longer paying panic prices for it. What would prove this wrong: a quiet expiry today, alerts lifted, maintenance season finishing without another emergency order, and December's capacity auction — where plants are paid simply to promise they will be ready in 2029 — clearing well below its $325-per-megawatt-day price cap. If October was weather luck, those things happen. If it was structure, they do not.

What the order actually authorized — the document, not the headline. Order No. 202-26-50 was issued Oct 12 by the Department of Energy under Section 202(c) of the Federal Power Act, on PJM's application of Oct 11, and runs Oct 13–16 (DOE order page, Tier 1). It does three specific things. First, PJM must dispatch the generating units named in the order's Exhibit A list and order them to run as needed for reliability — with temporary relief from environmental permit restrictions for those units, per PJM's own summary of the order. Second, only as a last resort — before declaring, or during, an Energy Emergency Alert 3 (EEA 3, the level at which firm load is interrupted) — PJM may direct backup generation resources to operate: auxiliary, standby, directly connected or battery storage, whether synchronized to the grid or not. Third, PJM may call on its transmission owners and distribution companies to implement it. The list was not static: PJM filed updated Exhibit A lists on Oct 12, Oct 13 and Oct 14 (DOE docket page), revising which units were covered while the emergency was already running. Secretary of Energy Chris Wright's stated basis was a statutory emergency — a sudden increase in demand and a shortage of energy or of generation and transmission facilities (DOE, Tier 1).

And what actually happened, in PJM's own record. The system never escalated past EEA 1 — PJM's lowest emergency designation — and no EEA 3 was declared, so the backup-generation directive remained authority on paper, not a reported use (PJM emergency postings, as reported Oct 14). PJM's operating updates (PJM Inside Lines, Tier 1 operator) fill in the week: generation outages were running at approximately 61,000 MW on Tuesday; PJM recalled or canceled transmission and generation outages where possible, including recall of all active generation maintenance outage tickets effective noon on Oct 14; and at 2 p.m. Tuesday its forecasts stood at 112,447 MW for Tuesday and 113,312 MW for Wednesday — later vintages of the same squeeze the application described. On Oct 14 PJM activated demand response: longer-duration resources across 18 zones, then a faster 30-minute program in the Baltimore Gas & Electric, Dominion and Pepco zones, the three territories the application named as most acute (PJM emergency messages, as reported). PJM's senior vice president for system operations, Mike Bryson, put the structural point plainly: “This is the new reality that grid operators are facing” (PJM, Tier 1). What would prove the emergency was a one-off: no extension filing, no new 202(c) application through the winter maintenance tail, and alerts standing down with today's expiry. What would prove it is the new operating mode: a fresh application at the next warm spell — in any month, not just summer.

Module 2 — The Macro Vector: A Thinner Cushion Underneath

While the electricity drama played out, the fuel underneath it quietly tightened. Thursday's EIA storage report showed 3,567 Bcf of gas in storage after a 67 Bcf weekly build — below the roughly 90 Bcf five-year average build for the week and below what traders expected. The surplus over the five-year average has now shrunk for weeks in a row, from 68 Bcf to 45 Bcf, and storage is 143 Bcf below last year. November gas futures rose about 2% on Thursday to $3.26 per million British thermal units (MMBtu, the standard pricing unit for gas) in response.

Why it matters in plain terms: stored gas is the shock absorber for winter. Every week the cushion shrinks before heating season even starts, the price of a January cold snap gets more expensive. The official outlook still helps the calm case — the EIA's October forecast expects storage to reach 3,850 Bcf by the end of October, about 2% above the five-year average — but recent builds have run below the pace that forecast assumes. Both things can be true: no shortage today, and less room for error than a month ago.

The regional detail inside the EIA report (EIA Weekly Natural Gas Storage Report, Oct 15, Tier 1, regional table as reproduced from the release) shows where the build did and did not land. The Midwest added the most, +27 Bcf to 1,037 Bcf (+2.0% versus its five-year average); the East added +22 Bcf to 884 Bcf (+3.3%); the Mountain region added +2 Bcf to 255 Bcf (+5.4%); and the Pacific actually drew down, −5 Bcf to 294 Bcf, while still sitting +6.5% above its five-year average. The tight spot is the South Central — the Gulf Coast region that feeds the LNG export terminals: it added +21 Bcf to 1,097 Bcf (salt-cavern facilities +9 Bcf, nonsalt facilities +12 Bcf) yet remains the only region below its five-year average, at −3.1%. The regions sum exactly to the national 3,567 Bcf, and the weekly changes sum to the +67 Bcf headline. Against the other benchmarks the build missed everything at once: pre-report estimates centred on about 74 Bcf, the same week last year built 80 Bcf, and the five-year average build is about 90 Bcf (EIA data, as reported). A year ago stocks stood at 3,710 Bcf — the arithmetic behind the 143 Bcf year-on-year deficit. The falsifier for the thinning-cushion thesis is concrete: two consecutive weekly builds at or above the ~90 Bcf five-year pace would re-widen the surplus back above 68 Bcf and break the erosion pattern; another sub-70 Bcf print next Thursday extends it. One more timing fact worth holding: the report was released at 10:30 a.m. ET Thursday, and the storage trajectory — not any single week's sign — is what winter pricing is being set on.

McKinsey's Global Energy Perspective 2026 makes the same point at system level (Tier 3 context): it describes an era of energy expansion — global demand up 2.7% in 2025 with every major fuel growing at once — where grids, wires, and firm capacity, not cheap generation, set the speed limit. That is a consulting firm's framing, not a market price; this week's PJM week is what the framing looks like in practice.

Module 3 — Dispersion: Thursday Sorted What Wednesday Lumped

Wednesday, Washington's who-pays signal hit every power stock at once. Thursday, they diverged — with an important nuance: Vistra and Constellation both own running plants and dated uprates, yet Vistra rose 1.6% to $154.17 while Constellation eased 1.6% to $275.00, and Talen eased 1.3%. So Thursday was not a clean category split (plants vs. paper); it was company-by-company pricing returning after a one-headline week. The paper end of the trade — projects not yet built — bounced hardest off low levels: NuScale +5.5% to $6.15 and Fermi +7.4% to $3.39. Treat both moves as unproven until they survive a second week: Fermi's bounce arrived with no company news, in the same week filings showed a major holder selling shares on Oct 13 and Oct 14. A bounce on no news, while a big holder exits, is trading — not validation. A bounce is not a business model.

There is also a regulatory reason the running-plant end should be priced differently, and it landed the same week. Jefferies analysts, in a client note reported by Utility Dive (Tier 2), judged that FERC's rejection of PJM's curtailment plan likely keeps PJM capacity prices elevated for longer — a direct benefit to existing generators, the companies that already own the megawatts the system is short of. That is a mechanism, not a mood: while the who-gets-cut rules are unwritten, the capacity market keeps paying scarcity prices to incumbents, and the $325-per-megawatt-day cap remains the reference point into December's auction. Priced in: Constellation at $275.00 sits between its $228.63 52-week low and panic pricing — the market is paying for elevated-for-longer capacity revenue, not for a resolution. Falsifier: if a state-led curtailment framework or a PJM refile emerges before the December auction with credible dates attached, the elevated-for-longer premium should compress first in exactly the names that rose on it — watch whether CEG, VST and TLN start moving together again on regulatory headlines rather than apart on company ones.

The sturdier dispersion signal came Friday morning from the equipment side. Schneider Electric raised its 2026 targets for the second time this year — core profit (adjusted EBITDA, earnings before interest, taxes, and depreciation) growth now expected at 18–21% and sales growth at 12–14% — after third-quarter revenue of €11.46 billion beat the €11.13 billion analysts expected, with its chief executive citing triple-digit demand growth in data-center orders, Reuters reported. Schneider does not own power plants or argue about capacity prices. It sells the gear every side of the fight needs. When the arms dealer raises guidance twice in a year while the generals argue over the bill, the buildout itself is not slowing — only the argument about who pays for it is getting louder.

Module 4 — Tactical Asymmetry: What to Watch

The FERC order, in full — because the reasoning is the roadmap. The Oct 9 order (Docket ER26-3515-000; cited as 197 FERC ¶ 61,019) accepted two pieces of PJM's Aug 13 filing and rejected the rest as one linked package (FERC order, Tier 1, as summarised in Tier 2 reporting). Accepted, effective Oct 12: the Large Load definition — end-use load with a cumulative peak of 50 MW or more at a single electrical site, with all affiliated facilities within a one-mile radius counted as one site, so a developer cannot duck the threshold by clustering smaller buildings — and the Large Load Registry. FERC declined to limit the definition to data centers: manufacturing and other large loads are in. Under the registry, load-serving entities must report each large load's location, in-service dates, peak load, ramp schedule, telemetry and backup generation; the data is visible to state commissions, FERC, the Market Monitor and market participants under PJM's existing confidentiality rules, while the public gets only aggregates on PJM's website. Rejected: the Interim Resource Adequacy Service (IRAS), the bring-your-own-new-capacity (BYONC) rules, the requirement to designate new capacity, and the plan to exclude uncovered new large loads from the capacity demand curve from the 2029/30 delivery year — FERC found those elements inextricably linked to IRAS. The reasoning had three limbs: parts of the tariff appear to place obligations directly on retail customers and dictate which of them get curtailed, intruding on state retail jurisdiction; the proposed compensation framework appears to establish a retail rate; and PJM had not shown the treatment was not unduly discriminatory — it offered no sufficiently detailed evidence for treating load growth served the same way differently depending on whether a new large load was behind it. Commissioner David Rosner concurred, with no dissent recorded: “We need state regulators to establish clear rules for which customers should be curtailed first” (FERC concurrence, Tier 1). His concurrence adds the sequencing point that will matter in any refile: IRAS would have forced new large loads off before voluntary demand response — customers paid to volunteer — and, in his words as reported, that flips the paradigm; non-critical data centers should be curtailed before homes, hospitals and schools, but only after voluntary demand response is exhausted, and states should set the order. FERC also pointed PJM toward a different instrument — a voluntary, non-firm transmission service of the kind raised in its show-cause proceeding — rather than a mandatory curtailment service.

Today, Oct 16 — the expiry. Watch whether PJM's alerts lift as the Energy Department's order ends, and whether any extension is filed. A quiet expiry supports the "managed well" case; an extension says the margin is still live.

By Nov 9 — PJM's compliance filing (and what it is not). FERC gave PJM 30 days from Oct 9 to file tariff language that keeps the Large Load definition and the registry but strips out everything rejected — the “New Large Load” definition (rejected without prejudice), IRAS and BYONC. Day 30 falls on Sunday Nov 8, so the filing is due Monday Nov 9. That makes Nov 9 a housekeeping deadline, not a second bite: a refiled curtailment proposal would be a separate, optional filing under Section 205 of the Federal Power Act, reshaped around FERC's guidance — voluntary rather than mandatory, demand response first, deliverable capacity — and no date for it was set in the order as reported. The registry (sites of 50 MW or more, live since Oct 12) is the part that survived — the first real census of data-center load on the largest US grid. Read the Nov 9 filing for one thing: whether PJM's language merely complies, or signals the shape and timing of a refile. Silence on a refile, paired with no state proceeding opening before December, would falsify the “states move fast” version of this story.

Ohio and Pennsylvania — the new front line, already moving. The state-led sequel is not hypothetical in either state. In Ohio, the who-pays half is already law of the tariff: the Public Utilities Commission of Ohio approved AEP Ohio's data-center tariff in July 2025, in effect since July 23, 2025 — new data centers of 25 MW or more must pay for at least 85% of their contracted capacity for up to 12 years (a four-year ramp, then eight years), regardless of actual use, with collateral and exit-fee provisions (PUCO order, Tier 1, as reported by POWER Magazine, Tier 2). That decision remains on appeal at the Ohio Supreme Court, per a Consumers' Counsel filing as reported, and Duke Energy Ohio has proposed its own Rate DC built on the same AEP framework, with contract-by-contract terms. What Ohio has not yet written is the who-gets-cut half — curtailment priority — which is precisely the piece FERC just handed to the states. In Pennsylvania, both halves are in motion: on Oct 1 the Public Utility Commission advanced a tentative emergency load-control order whose draft specifies that data centers are not critical load (RTO Insider, Tier 2); and Governor Shapiro's Executive Order 2026-05 (signed Aug 18) already directs the administration's Special Counsel for Energy Affordability to press the PUC to curtail data centers before other customers in certain emergencies unless they have secured incremental capacity for their full demand, to deny them critical-load exemptions, and to assign them interconnection and PJM reliability costs (Executive Order 2026-05, Tier 1, as reported). The order's own recitals put Pennsylvania's proposed-facility count at more than 100 data-center projects, against permit applications received for about 20 — the state's own five-to-one discount on announced load, as summarised from the order. Pennsylvania also supported PJM's curtailment approach before FERC, per Latitude Media's reporting — Harrisburg wanted the federal version and will now have to write its own. What to watch, in order: whether the PA PUC's tentative order becomes final and on what timeline; whether Ohio opens a curtailment-priority proceeding or waits for its Supreme Court appeal; and which state's rules land first, because the first final order becomes the template the other twelve PJM states copy. State hearings are now market-moving events for every plant owner in PJM.

December — the capacity auction for 2029/30. The last auction (2028/29 delivery year, July 2026) cleared 6,831 MW short of the reliability requirement — the reserve cushion PJM is required to hold — at the $325 cap, the second straight year the auction has come up short, as FERC's own order acknowledges in recognising “the imminent capacity shortfall in PJM.” The December auction runs under the same collar — a $325 cap and a $175 floor, extended through 2029/30 (PJM, Tier 1, as reported) — and one escape valve is closed in the meantime: the separate backstop procurement for up to 6,831 MW is suspended to Feb 28, 2027, as reported from the FERC record. A repeat print at the cap confirms scarcity pricing; a clear drop below it would be the first hard evidence the squeeze is easing.

The numbers, in one place (Thursday closes unless noted): Constellation $275.00 (−1.6%), Vistra $154.17 (+1.6%), Talen $353.60 (−1.3%), NuScale $6.15 (+5.5%), Fermi $3.39 (+7.4%); PJM spot $1,021 per megawatt-hour Tuesday, next-day $295.83 (+28%, Bloomberg); capacity cap $325 per megawatt-day; gas storage 3,567 Bcf (+67 Bcf, EIA); November gas $3.26/MMBtu.

Balance of risks: owners of running plants keep the strongest hand while capacity prices sit at the cap — the risk to them is political (bills), not physical. Paper projects carry the opposite profile: any bounce can reverse on financing or insider selling. Equipment sellers sit in the middle, exposed to the buildout rather than to any single ruling.

Module 5 — Value Creation vs Value Capture: The Grid Held. Price the Holding.

Value creation this week was unglamorous and real: dispatchers who kept 67 million people's power on, customers who cut use when asked, maintenance crews called back to work, and engineers squeezing 433 MW and 890 MW more out of reactors that already exist. Value capture was the louder trade: shortage prices, a $325 capacity cap doing the work of a business plan, and paper projects swinging down 15.1% on Wednesday and up 7.4% on Thursday on trading flows alone. The honest lesson of the week is that creation quietly won — the system worked — but it won using emergency tools that were designed to be rare. An emergency order that expires on schedule is a success story. A grid that needs one every season is a business model — somebody else's. Watch who gets paid for the holding, and who is actually doing it.

Sources

• tED Magazine — DOE Order No. 202-26-50, Oct 12, 2026, effective Oct 13–16; PJM forecast demand ~111,500 MW, within ~900 MW of reserve margin Oct 14, ~61,591 MW of anticipated outages. Summarizing the DOE order. Tier 1 originator summarized; order via DOE.
• TradingView (Seeking Alpha) — PJM spot $1,021/MWh Tuesday; day-ahead Wednesday +28% to $295.83/MWh, Bloomberg reported; PJM demand expectation 113,312 MW Oct 14. Tier 3 carrying Tier 2 Bloomberg reporting.
• Bloomberg — PJM spot and day-ahead prices; administration/states who-pays plan (Oct 14–15 reporting). Cited by name; Bloomberg article URL not retrieved. Tier 2.
• SRN News — Reuters, Oct 16: Schneider Electric raises 2026 targets (adjusted EBITDA (core profit) +18–21%, organic revenue +12–14%), Q3 revenue €11.46B vs €11.13B consensus, triple-digit data-center demand growth. Tier 2 (Reuters).
• Capital Market via HT Syndication — EIA Weekly Natural Gas Storage Report, Oct 15: working gas 3,567 Bcf for week ending Oct 9 (+67 Bcf), 143 Bcf below last year, 45 Bcf above five-year average of 3,522 Bcf. Tier 1 (EIA) as reported.
• U.S. Energy Information Administration — Short-Term Energy Outlook (Oct 6): end-September stocks 3,523 Bcf; forecast 3,850 Bcf at end-October, ~2% above five-year average; Henry Hub 2026 average forecast $3.48/MMBtu. Tier 1.
• TradingNews — November Henry Hub futures $3.26/MMBtu, +~2% Thursday Oct 15 after the storage report; build below 74 Bcf estimate and 90 Bcf five-year average build. Tier 3.
• Transformer Magazine — Source: Reuters: FERC approves PJM data-center registry, rejects temporary-disconnection proposal; PJM forecasts data-center demand to triple by 2030 to ~29 GW; 6.8 GW capacity shortfall noted. Tier 2 (Reuters) as carried.
• The Duck Curve — FERC order Oct 9 (accession 20261009-3055): Large Load definition (≥50 MW at a single site) and Registry accepted, effective Oct 12; the interim curtailment service (IRAS) and bring-your-own-capacity (BYONC) rules rejected. Tier 3 summarizing Tier 1 FERC order.
• The Daily Energy — Jefferies view via Utility Dive (Oct 12): ruling likely keeps PJM capacity prices elevated longer, favoring existing generators; states (Ohio, Pennsylvania) likely to act first. Tier 3.
• Finnhub market data — Oct 15 closes: CEG $275.00 (−1.62%), VST $154.17 (+1.57%), TLN $353.60 (−1.34%), SMR $6.15 (+5.49%), FRMI $3.39 (+7.44%); CEG 52-week low $228.63. Wednesday Oct 14 FRMI close $3.16 (−15.05%) per Finnhub, as reported in Issue #7. Tier 3.
• Ticker Report — SEC filing: Fermi major shareholder sold 2,750,000 shares on Oct 13 at $3.73 average. Tier 3 reporting Tier 1 SEC filing.
• FXEmpire — EIA report for week ending Oct 2: storage 3,500 Bcf, 68 Bcf above the five-year average of 3,432 Bcf (the prior-week cushion compared with this week's 45 Bcf). Tier 3 reporting Tier 1 EIA data.
• Insider Monkey — Constellation $4.3B uprate program: 890 MW across 11 reactors, first upgrade expected by 2028; Vogtle cost comparison. Tier 3.
• The Cerbat Gem — SEC filing: Fermi major shareholder sold 1,297,200 shares Oct 14 at $3.15 average; FRMI traded up $0.23 Thursday to $3.39. Tier 3 reporting Tier 1 SEC filing.
• Motley Fool — DOE conditional loan up to $4.2B for Vistra nuclear upgrades; Google 20-year deal for 890 MW of Constellation uprates backed by $4.3B investment. Tier 3.
• Construction Review Online — Vistra uprate program factsheet: up to $4.2B conditional loan; +433 MW across Perry, Davis-Besse, Beaver Valley; ~4 GW preserved. Tier 3.
• McKinsey & Company — Electrification equipment trends: data centers could exceed 5% of electricity consumption and 20%+ of demand growth through 2030; grid connections and supply chains as local constraints. Tier 3.
• McKinsey & Company — Global Energy Perspective 2026 (Sep 30): global energy demand +2.7% in 2025; data centers fastest-growing OECD load segment, 24% CAGR to 2030 under Continued Momentum; grid investment framing. Cited by name; full McKinsey report URL not linked here. Tier 3.
• PJM Base Residual Auction, 2028/29 delivery year (July 14, 2026) — cleared 6,831 MW short of the reliability requirement at the $325/MW-day cap. Via PJM auction results as reported. Tier 1.
• U.S. Department of Energy — Order No. 202-26-50 page: issued Oct 12, 2026 under Federal Power Act Section 202(c) on PJM's Oct 11 application; effective Oct 13–16; dispatch of specified units, backup generation as a last resort before or during EEA 3; PJM Exhibit A updates filed Oct 12, 13 and 14. Tier 1 (primary document).
• PJM Inside Lines — PJM operations update: Maximum Generation Alert extended, demand response called, generation outages approximately 61,000 MW on Tuesday, recall of all active generation maintenance outage tickets effective noon Oct 14, load forecasts 112,447 MW (Tue) and 113,312 MW (Wed) as of 2 p.m. Tuesday; Mike Bryson statement. Companion update: Maximum Generation Alert for Oct. 13; environmental-permit relief and backup-generation authority under the DOE order. Tier 1 (grid operator).
• U.S. Energy Information Administration — Weekly Natural Gas Storage Report, Oct 15, 2026 (week ending Oct 9): working gas 3,567 Bcf (+67 Bcf); regional stocks East 884 Bcf (+22), Midwest 1,037 Bcf (+27), Mountain 255 Bcf (+2), Pacific 294 Bcf (−5), South Central 1,097 Bcf (+21; salt +9, nonsalt +12); South Central the only region below its five-year average (−3.1%). Regional table as reproduced from the EIA release. Tier 1 (primary).
• Headwaters Wire — FERC order, Oct 9, Docket ER26-3515-000: Large Load definition and Registry accepted effective Oct 12; IRAS, BYONC and demand-curve exclusion rejected as linked; retail-jurisdiction and undue-discrimination reasoning; Rosner concurrence. Order cited as 197 FERC ¶ 61,019 in FERC-record summaries. Tier 2 summarising Tier 1 FERC order.
• Utility Dive — FERC rejects most of PJM's large-load plan, Oct 12: Jefferies view that the ruling likely keeps PJM capacity prices elevated longer, favouring existing generators, with any revival likely state-led (Ohio, Pennsylvania first); Rosner concurrence detail on curtailment sequencing after voluntary demand response. Tier 2.
• POWER Magazine — Pennsylvania Executive Order 2026-05 (Aug 18): GRID requirements for data centers above 25 MW; direction to press the PA PUC on curtailment before other customers absent full incremental capacity, no critical-load exemption, and cost assignment. Tier 2 reporting Tier 1 executive order.
• RTO Insider — Pennsylvania PUC advances large-load emergency curtailment plan, Oct 1: tentative order specifies data centers are not critical load. Tier 2.
• POWER Magazine — PUCO adopts AEP Ohio data-center tariff (July 9, 2025; in effect July 23, 2025): new data centers above 25 MW pay at least 85% of contracted capacity for up to 12 years (four-year ramp), with exit fees and financial assurance. Tier 2 reporting Tier 1 PUCO order.