This week: two things happened four days apart that together reset the demand side’s argument. On October 9 — reported October 12 — FERC rejected the core of PJM’s large-load interconnection plan, approving only the 50-MW large-load definition and a large-load registry while striking the Interim Resource Adequacy Service (IRAS) and the bring-your-own-capacity (BYOC) rules on jurisdictional grounds. Commissioner Rosner’s concurrence held that curtailment should come “only when necessary for reliability, after voluntary demand response is used up,” and that states — not FERC — should set the order after that. Then on October 13 PJM did exactly that: with roughly 61,000 MW of generation offline for autumn maintenance and forecast load near 112,447 MW Tuesday and 113,312 MW Wednesday, it issued a Maximum Generation Alert and a Pre-Emergency Load Management Action calling on demand response, backed by a fresh DOE Section 202(c) order (No. 202-26-50, effective October 13–16) — the first §202(c) order this series has recorded in a shoulder-season month rather than a summer or winter peak.
Put those together and the week’s question is what is actually dispatchable now. The federal mechanism that would have ranked large loads in the curtailment stack was struck down and remanded, in effect, to fifty state proceedings. Every supply-side answer on the board this week is late: PJM’s first expedited-interconnection approvals (2.1 GW, roughly 80% of it batteries) are due mid-2029; SoftBank’s ~10-GW Ohio complex is phased through 2032; Wood Mackenzie has gas turbines headed to $600/kW by end-2027 against 35–116 GW of OEM backlog. Voluntary demand response is the one instrument that answered a call in October 2026, and Brattle put a number on how much of it is sitting unenrolled: 10 GW → 35 GW of residential VPP capability at a 4% system-peak reduction, worth $4–7B a year.
Last week’s argument was about who gets credit for flexibility. This week’s is narrower and stronger: flexibility was the only resource that arrived on time, and the rules for ranking it just moved from one FERC docket into every state commission’s jurisdiction. That is both the opportunity and the compliance problem of the quarter.
⚡ Virtual Power Plants & Demand Flexibility
PJM called demand response in mid-October, and that is the single most useful fact in this week’s log. (deduplicated — reported across four sources in the October 14 entry) On October 13, PJM issued a Pre-Emergency Load Management Action alongside a Maximum Generation Alert and a Load Management Alert, to cover the afternoon and evening net-load peaks — the hours when solar output falls to zero while demand stays high. The operator also recalled non-firm exports to neighboring grids and recalled all active generation-maintenance outage tickets effective noon October 14. The setup: approximately 61,591 MW of generation already offline for autumn maintenance, forecast demand of roughly 111,500 MW on October 13, and a margin that came within about 900 MW of the reserve requirement on October 14. DOE issued Section 202(c) Order No. 202-26-50 on PJM’s October 11 application, effective October 13–16, directing which units must run and authorizing backup generation as a last resort before or during an Energy Emergency Alert Level 3, across PJM’s 13 states plus the District of Columbia.
Three details matter for a business case rather than a news summary.
First, the order of operations held in practice. PJM reached for pre-committed, voluntary demand response first, and reported that it had not yet needed large-load backup generation — which is precisely the sequence Rosner’s concurrence described four days earlier. A utility arguing for a verified DR portfolio now has an operational instance, not just a tariff theory, and it is in the same month as the ruling that made voluntary DR the first-called resource.
Second, the order reaches distributed assets on terms worth reading closely. Order No. 202-26-50 covers auxiliary, standby, directly connected and battery-storage resources whether or not they are synchronized to the bulk power system, and explicitly contemplates data centers and other large loads running their own backup fleets to reduce grid reliance. Critical facilities — hospitals, 911 centers, water utilities, defense sites — are carved out. Environmental relief is narrow rather than a blanket waiver: affected units may run only at the times and parameters PJM deems necessary, must remain consistent with environmental requirements “to the maximum extent practicable,” and stay under daily reporting and recordkeeping, with normal law resuming at expiry. Critically, the order does not convert any of these resources into capacity-market participants. It is an emergency tool, not a standing DER program — which means it confers none of the accreditation that would let those megawatts count as firm capacity in an IRP.
Third, DOE itself articulated the latent-capacity thesis. In its own announcement, the department estimated more than 35 GW of unused backup generation sits idle nationwide and framed it as a reliability reserve, with Secretary Chris Wright describing the order as making “additional power generation, including backup power, available to use as needed.” The policy irony is the operative argument: the same 35 GW that DOE wants available by emergency fiat is, in aggregate, exactly the fleet a standing VPP or DERMS program would meter, enroll and dispatch on a planned basis with performance accreditation attached. An emergency order is the expensive way to get a resource a program could have made firm.
The gap in the record is the number a filing would need: PJM published no deployed DR megawatts and no performance results for the October 13–16 window as of this writing, and a targeted search found none. Treat the event as qualitative evidence of dispatch priority and timing, not as a measured performance figure. (Sources: T&D World, October 13; PJM Inside Lines, October 13; Microgrid Media, October 12; U.S. DOE, October 12 — October 14 entry)
Brattle sized the unenrolled resource, and the benchmark converts cleanly into an IRP test. In a report dated October 13 prepared for aggregator Everyday Electric (the former Renew Home, whose rebrand this series reported last week), Brattle modeled a scenario in which every U.S. utility achieves a 4% system-peak reduction through residential virtual power plants — a level the top 10% of utility DR programs already achieve. Nationwide residential VPP capability would rise from about 10 GW to 35 GW, producing $4–7B per year in household benefits: roughly $1.8–5.5B shared across all ratepayers and about $1.9B paid directly to participants, enough to cut the average household’s bill increase over the next five years by more than half. The named lever is enrollment, specifically OEM-led enrollment that activates capability already sitting in installed thermostats, batteries and EV chargers rather than requiring new hardware. Brattle principal Ryan Hledik: “a significant amount of untapped flexible capacity already sitting in customers’ homes.” The near-term driver is stated bluntly — “recent doubling of natural gas-fired generation capacity costs is an acute near-term concern” — against U.S. residential rates up 33% nominal from 2019 to 2025.
The 4% figure is the one to carry, because it is a portfolio test rather than a national projection: 4% of system peak is roughly 40 MW of residential flexible capacity per GW of peak, which at this series’ standing Brattle avoided-capacity value of about $66/kW-year implies on the order of $2.6M/year of capacity value per GW of system peak before energy, T&D-deferral and ancillary stacking. Run your own portfolio against it; the distance between your number and 4% is the enrollment gap, stated in megawatts rather than adjectives.
Two caveats must travel with every citation. The study was commissioned by an aggregator that benefits from VPP growth, and the 35 GW is a uniform-attainment scenario, not a forecast — the gap between top-decile and median programs is exactly the enrollment and dispatch-reliability problem the scenario assumes away. Note also that this is a different Brattle study from the Uplight/SPP analysis reported October 2 (the one finding 23 of 90 MW from orchestration alone); the two are complementary, not a repeat, and the orchestration finding remains the more defensible line in a capital request because it does not depend on an enrollment forecast at all. (Source: Utility Dive, October 14 — October 15 entry)
Cornell proposed a market design that turns a large load’s curtailment obligation into demand for distributed capacity. Cornell engineers, in work published in Applied Energy on September 22 and publicized October 16, propose a Curtailment Credit Market layered on the non-firm interconnection agreements that let large loads connect faster in exchange for accepting curtailment. When the operator signals a constrained line, a data center facing a cutback can instead buy verified reductions from other eligible users — factories, large commercial customers, virtual power plants, or aggregators coordinating batteries, solar, EVs and building controls — with each trade cleared only if it is physically deliverable on the affected part of the network and actually produces the required reduction. Tested on three simulated grids, one a simplified version of New York’s, trading avoided at least a third of the economic damage of forced cutbacks versus curtailing the data center outright, using a hypothetical 100-MW reduction as the illustration. Lead author Richard Mahuze, overseen by Prof. Max Zhang, with Ali Amadeh and Charlotte Gressel contributing.
Why it belongs in a procurement file rather than a reading list: the design treats locationally verified load flexibility as a fungible, market-cleared product — the same thing a DERMS or VPP already meters, dispatches and settles — and it reframes non-firm interconnection as a demand channel for distributed capacity rather than a one-sided risk the connecting load absorbs alone. That is a sales motion, not an advocacy position: the counterparty is a large-load customer with an obligation to discharge, which is a far easier conversation than persuading a commission to value avoided capacity. It also lands in the same week FERC floated a voluntary non-firm transmission service (below), which is the regulatory half of the same idea. The caveats are explicit and should be stated in any citation: the result is modeled, not operational, the concept has not been piloted, and the deliverability-verification and settlement machinery a real program would require is untested. (Source: Cornell Chronicle, October 16 — October 16 entry)
PJM’s own September review supplies the operational record behind the shoulder-season thesis — and shows the October event was not an anomaly. PJM presented a review of two unseasonable September events to its Operating Committee on October 12. The August 31 – September 4 event peaked at about 152,800 MW at 5:30 p.m. on September 1, with generation outages peaking at 15,502 MW (8.47%). The September 15–18 event peaked just under 127,000 MW at 4:50 p.m. on September 17, with outages at 37,916 MW — 20.7% of system capacity — on September 18. In both, PJM called targeted demand response: on September 1 through the net-load peak, where it also relieved transmission constraints that were bottling generation, and earlier in the operating day during September 15–18. A DOE 202(c) order effective September 2–8 authorized PJM to direct large loads to backup generation. Exports to equally hot neighbors and congestion in the D.C./Maryland/Virginia zones added stress.
PJM’s own stated lesson is the one that belongs in an IRP: shoulder-season maintenance “can’t simply [be] defer[red] indefinitely,” and rising demand is outpacing new generation additions, tightening conditions in exactly the months planners have treated as slack. For DR program design the implication is concrete and testable: dispatch windows, event-hour assumptions and capacity accreditation built around summer afternoons systematically undercount resources available at 5–8 p.m. in September and October. Three events in seven weeks — September 1, September 17, October 13 — all in the post-sunset net-load ramp, is a pattern, not a run of bad luck. As with October, the gap is that PJM published no DR megawatts and no performance results. (Source: PJM Inside Lines, October 12 — October 15 entry)
Movement — PJM’s reliability-backstop procurement quietly became more workable for aggregators. This series reported FERC’s acceptance-and-suspension of PJM’s 6.8-GW backstop on October 2; a Latitude Media analysis from October 1 that entered the log this week adds the DER-specific mechanics those reports omitted. FERC accepted the broad framework but delayed the effective date five months and remanded the centralized-auction portion — which Jefferies called an “unexpected twist” that extends the window for bilateral deals on distributed capacity and could shrink how much PJM must buy at auction. That matters because, as Aurora Energy Research’s Julia Hoos notes, the auction’s requirement for exact asset locations and 15-year availability guarantees fits demand response poorly — an aggregator’s customer mix shifts year to year — whereas the bilateral phase is far more workable.
Two eligibility concessions are the actionable part: PJM agreed to let previously DR-enrolled sites that sat out the prior year count as “new” capacity, letting aggregators reactivate dormant locations, and to count incremental capacity such as new batteries added at existing DR sites. Both are direct wins for behind-the-meter storage and VPP build-out, and both are the kind of definitional detail that decides whether a portfolio qualifies. The caveat that keeps this a mixed result: FERC has not confirmed whether distributed capacity offsets a large load’s IRAS obligation — and with IRAS now rejected outright, that demand channel is more uncertain than it was, so near-term aggregator value runs through the bilateral backstop, not the interconnection track. (Source: Latitude Media, October 1 — October 13 entry)
📋 Regulatory & Policy
FERC rejected the core of PJM’s large-load interconnection plan, and the grounds matter more than the outcome. (deduplicated — reported across three sources in the October 13 and October 16 entries) In an order issued October 9, 2026 in Docket ER26-3515-000 — the docket this series has been chasing for nine weeks with no public record — FERC approved PJM’s 50-MW single-site large-load definition and a large-load registry recording each facility’s location, ramp schedule and capacity supply, while rejecting the two provisions that carried the policy: the Interim Resource Adequacy Service (IRAS), which would have let large loads connect without matching new capacity if they accepted curtailment ahead of other pre-emergency demand response, and the bring-your-own-capacity (BYOC) rules. PJM had filed on August 13 seeking an October 12 effective date, applicable to large loads entering service after June 1, 2027.
The rejection was jurisdictional, not conceptual. FERC found that several provisions appeared to impose obligations on new large loads that are mostly retail customers, to dictate which retail customers get curtailed, and to set a retail compensation rate — all of which intrude on state authority and may be discriminatory. FERC separately found PJM’s capacity-accreditation lock-in unsupported. The analytical follow-up adds two specific BYOC objections a policy summary misses: FERC objected to locking in accreditation values for 10 years, and to letting BYOC capacity be delivered anywhere in the market rather than in the constrained zone where the load physically sits. FERC also floated a voluntary non-firm transmission service, in which flexible loads accept curtailment risk in exchange for avoiding firm-capacity costs and getting faster interconnection.
Four consequences for demand-side practitioners, in order of how soon they bite.
1. Voluntary DR is now first in the emergency dispatch order, on the record. Rosner’s concurrence states that non-critical data centers should be curtailed before homes, hospitals and schools — but that states should set those priorities, and only after voluntary demand response is exhausted. Removing IRAS takes away the one mechanism that would have made large loads curtailable ahead of conventional demand response, which leaves aggregated DR and behind-the-meter resources relatively stronger in the emergency stack. PJM’s October 13 action is the first instance of that sequence running in the field.
2. The curtailment-order fight moved to the states, which fragments the rulebook. Analysts read the order as an invitation to an optional new Section 205 refiling, likely state-driven, with Ohio and Pennsylvania named. Aggregators are relatively unconcerned for exactly this reason: Voltus expects states to stand up their own IRAS-style regimes, and Mary Yang of 804 Advisory (a former PJM regulatory counsel) reads the ruling as a setback, not a dealbreaker, with many BYOC deals already queued to close and some buyers likely to look at other regions while PJM decides whether to refile. Pennsylvania Governor Shapiro had backed IRAS. The pointed open question is cross-state: when a single transmission zone spans two states, who sets the curtailment order? For a multi-jurisdictional utility, state-by-state divergence in DR and DSM program rules is now a planning assumption rather than a risk.
This is the direct continuation of a thread this series has carried since August. Pennsylvania’s October 1 tentative order — EDCs must shed new large loads without secured capacity before existing residential and commercial customers, with data centers barred from “critical load” status — is now not just an early mover but the template FERC pointed the states toward, four days after the fact. A state commission’s fastest route to a flexibility outcome is its own jurisdiction, and PA has the only finished product. Its comment clock runs 30 days from publication in the Pennsylvania Bulletin plus 15 days reply, with a cost-allocation technical conference November 17 and a final order targeted January 28, 2027. The Bulletin publication date — which sets the actual deadline — remains unverified as of this writing, and it is the single most time-sensitive item on this board.
3. Deliverability-in-zone, if BYOC returns, would reprice locational distributed capacity. FERC’s objection to delivery “anywhere in the market” is a standard that, if carried into a refiling, rewards locationally sited DER and storage over generic market capacity — which is the premium a DERMS-managed, feeder-aware portfolio exists to capture. Track it as a design input, not a rhetorical win.
4. The capacity-price signal cuts toward the demand side. Rosner flagged that excluding IRAS loads from the demand curve could make PJM look over-supplied while it sheds load involuntarily. Jefferies expects the ruling to keep capacity prices higher for longer (naming Constellation, NRG, Talen and Vistra as beneficiaries) — which raises the avoided-capacity value that DR and EE programs can claim in IRP filings. Meanwhile both of PJM’s answers to the 6,831-MW 2028/29 shortfall are now stalled: IRAS rejected, backstop procurement suspended five months over cost allocation, against a 138,318 MW procured in the last Base Residual Auction. For contrast, Texas set its large-load threshold at 75 MW and ERCOT sits outside FERC wholesale jurisdiction entirely — so the 50-MW convergence this series has tracked now has a federal anchor in PJM and an explicit divergence next door. (Sources: Utility Dive, October 12; Mgrid.org, October 10; Latitude Media, October 13 — October 13 and October 16 entries)
Movement — DOE filed at FERC for the first time in at least five years, and it filed on cost causation. On October 7, the U.S. Department of Energy filed a statement of position at FERC — reportedly its first FERC filing in at least five years — urging PJM to revise its reliability-backstop procurement so that data centers and other large loads, not households, fund the generation and infrastructure they require. The filing reinforces FERC’s September finding that parts of the plan may be unjust and unreasonable on cost allocation, transmission-owner exit rules and load-serving-entity collateral requirements, which paused the plan’s September 30 start. PJM plans a revised filing by October 29 and held a special Members Committee meeting October 22. DOE argues costs should be allocated using updated load forecasts, and that PJM needs project-level data to avoid double-counting, reconcile its base forecast, adjust targets when forecast load fails to materialize, and assign demand growth to the correct zone and LSE. DOE cites the White House Ratepayer Protection Pledge, whose PJM signatories include AEP, AES, CenterPoint, Dominion, Exelon, FirstEnergy and PPL.
The DSM through-line is twofold and both halves are useful. The 6.8-GW shortfall PJM is trying to cover — after two consecutive base auctions cleared at the price cap — is precisely the gap cost-effective demand-side resources can shrink. And the push for project-level large-load visibility is the same granular data foundation that flexible-load and curtailable-service programs depend on; a utility asking for that telemetry now has a federal filing to cite. More broadly: overstated or double-counted large-load forecasts inflate the avoided-cost case for supply-side build, and disciplined project-level forecasting makes the demand-side alternative look better by comparison. This series has tracked the same arithmetic through AEP Ohio’s 30 GW → 13 GW revision and Exelon’s ~40% haircut; DOE has now adopted the method. (Source: Utility Dive, October 8 — October 12 entry)
🏗️ Data Centers & Large Load Growth
A single 10-GW project is now moving through a bespoke state mechanism, and it is the counterfactual the demand-side case is argued against. PJM filed a study agreement at FERC on October 13 (retroactive to a September 25 execution) to evaluate SoftBank’s roughly 10-GW “Flagship” data-center complex, announced in February 2026 as part of a $550B U.S.–Japan trade deal with an estimated $33B for 9.2 GW of gas. An AEP Ohio data-center tariff would serve an initial 800-MW phase, while the PJM study covers an added 9.6 GW of load and 10.2 GW of new SB Energy gas generation phased through 2032, plus three AEP Ohio 765-kV lines — most of it sited at the DOE’s Portsmouth Gaseous Diffusion Plant in Pike County.
The novel feature is the channel. Ohio is using PJM’s State Agreement Approach, under which a state asks PJM to study projects needed for state policy and the requesting state’s consumers pay. PJM’s study informs Ohio’s decision only — it does not select projects, set cost allocation, or grant interconnection rights. SB Energy says it will fund the generation and transmission built for the data centers, with infrastructure payments that apply regardless of how much power the complex uses, and commits to making excess capacity available to the grid. Final cost allocation still runs through PUCO and FERC.
Two things to hold onto. First, the ratepayer-protection structure — consumption-independent payments — is the strongest form of the commitment instrument this series has tracked across 100-plus large-load tariffs, and interveners will test whether it survives contact with a cost-allocation proceeding. Second, and more useful for a DSM filing: a single 10-GW, build-everything project financed off a bespoke state mechanism and phased through 2032 is the slow, capital-heavy alternative that a verified DR/VPP portfolio and load flexibility are meant to defer or shrink. Price your demand-side option against it the way this series has recommended since October 2 — as cost per kW available in 2028, not per kW-year. Nothing in the Flagship structure delivers a megawatt before the window where the adequacy gap actually binds. (Source: Utility Dive, October 14 — October 16 entry)
Michigan has two approved GW-scale data-center contracts with no reported flexibility terms at all, in a state that just watched PJM call DR. The Michigan PSC has approved two data-center contracts — Oracle’s 1.4-GW campus in Saline Township and Google in Van Buren Township — both requiring the customer to pay for its own infrastructure upgrades. The Saline project alone is expected to raise DTE Energy’s load by about 25%, and DTE reports roughly 6 GW more in its data-center pipeline beyond its OpenAI and Google agreements. On the cost side, Consumers Energy CEO Garrick Rochow claims an average residential customer could see about “$7.50 per month of bill benefit with every gigawatt of new large load” — a utility claim, not an independent finding — and Indiana Michigan Power has proposed rate reductions tied to data-center growth. Independent experts quoted say data centers are not raising rates “at least for now,” but that rates “could rise” if grid-upgrade costs end up shared. The process risk is live: Attorney General Dana Nessel and environmental groups are challenging the Saline approval as “fast-tracked.”
The 25% figure is the operative number — a single contract moving a utility’s load by a quarter is the kind of step-change that reorders resource-adequacy timelines and raises the capacity value of every flexible megawatt, including the large customer’s own. And the gap in the reporting is itself the finding: two approved gigawatt-scale contracts, neither with reported curtailment, flexibility or demand-response terms, in a state inside PJM’s footprint that was under a Maximum Generation Alert the same week. Michigan is also the state where PBR momentum is building (PSC Chair Dan Scripps‘ affordability recommendation, SB 768 extending the rate-case interval from one to three years, residential rates up 10.42% YoY against 4.98% nationally). Those two threads belong in the same filing: a large-load contract with no flexibility obligation is a forgone flexibility resource, and under PBR it is also a forgone earnings opportunity. (Source: Bridge Michigan, October 14 — October 15 entry)
PJM’s first fast-track interconnection approvals set the supply-side benchmark the demand side must beat — and it is 2029. On October 8, PJM approved its first three projects under the Expedited Interconnection Track (EIT), totaling 2.1 GW: two Engie battery-storage projects (860 MW in Crawford, Pennsylvania and 800 MW in Morrow, Ohio — nearly 1.7 GW combined) and a 455-MW uprate of LS Power’s 810-MW Hunterstown combined-cycle gas plant in Adams, Pennsylvania. All are expected online by mid-2029. FERC approved the EIT in June 2026; PJM opened applications July 31. Eligible projects must exceed 250 MW, have primary siting-authority backing, and be able to operate within three years; PJM will consider up to 10 requests per year, and the process sunsets at the end of 2027.
The comparative read is the point. Every gigawatt PJM rushes onto the system through expedited generation interconnection is a gigawatt that does not have to come from the backstop procurement or from demand-side resources — so the speed and cost of these options set the benchmark a demand-side value stack has to beat. Two observations sharpen it. Roughly 80% of the approved megawatts are battery storage, which both confirms that the resource filling the near-term gap is increasingly four-hour storage and reprices the avoided-capacity value DR and EE can claim against it. And mid-2029 is the fast track — which means for the 2027 and 2028 gap years, there is no supply-side answer in this queue at all. That is the window where demand-side resources are not merely cheaper but structurally unopposed. (Source: Utility Dive, October 9 — October 12 entry)
🔌 DERMS & Grid Integration Technology
Interconnection-study throughput is the chokepoint behind every DER value stack, and the proposed fix is a planning-software problem. In an October 15 op-ed, Joseph Santamaria, AWS’s general manager for energy and utilities, argues that the interconnection queue is a study-throughput problem rather than a project-shortage problem: more than 2,000 GW of potential capacity awaits review, while the average interconnection wait has stretched from about two years in the early 2000s to roughly five years and rising, against demand projected to grow more than 50% by 2050. His remedy is AI planning tools that model thousands of what-if scenarios at once rather than one at a time, with human engineers making final decisions. The cited results: Duke Energy doubled the number of projects it evaluates in a year; PG&E combined data from 80-plus sources across 20 systems to prioritize 17,000-plus upgrades, cutting costs 20% in a $100M pilot and eliminating millions of customer-outage minutes; and an Amazon–NIPSCO agreement is expected to save existing customers about $1B over 15 years.
For the DERMS/ADMS thread the relevance is adjacency, not identity: these are planning and interconnection-study tools, not distribution control systems. But the same data-integration and simulation layer that unclogs a transmission queue is what a utility needs to run hosting-capacity analysis and locational DER dispatch — and a faster queue directly improves the economics of distributed and demand-side resources, because every month of queue delay is a month the deferral value cannot be booked. The caveats are substantial and should be stated whenever this is cited: it is an AWS-executive opinion piece, the results are the author’s account of utility and partner projects with no independent verification, and no grid-software vendor or product is named. EPRI research is cited in the piece, and FERC convened AI-for-grid-planning discussions over the summer. (Source: Utility Dive, October 15 — October 16 entry)
A joint-action agency just set the distribution asset-data terms for 32 utilities at once — which is how small public power will eventually buy a DERMS. Noteworthy AI was selected through the Florida Municipal Power Agency’s competitive RFP for an “AI Utility Infrastructure Data Collection System,” so each of FMPA’s 32 member utilities can now buy on standardized terms and negotiated group pricing through FMPA’s Joint Purchasing Program. FMPA’s members serve more than 2.5 million residents, roughly 11% of Florida. The service mounts AI-enabled cameras on vehicles to geolocate poles and capture timestamped imagery, with cloud analysis for asset inventory and GIS development, pole and equipment condition, joint-use attachment inventories, lighting audits, vegetation detection and post-storm damage assessment. Members can buy cameras for fleet vehicles, rent them for targeted patrols or storms, or hire turnkey collection. The one named result: the City of Starke, where about eight hours of collection inventoried roughly 2,300 poles and a follow-on audit found previously uncounted pole-attachment revenue.
This is a vendor release and is rated accordingly, but it earns its slot for a structural reason this series keeps returning to: an accurate distribution network model — pole, transformer and connectivity data in the GIS — is the prerequisite for any DERMS hosting-capacity calculation or ADMS switching function, and small public power utilities are the least likely to have one. The channel is the finding: a joint-action agency running one RFP for 32 utilities is how that cohort will acquire the data foundation and, later, the DERMS itself — which means GIS data-quality and connectivity-model accuracy requirements are likely to be written at the agency level, not utility by utility. If you sell to or benchmark against public power, the agency is the specification venue. The caveats are significant: the “up to 80% lower” cost claim and all benefits are vendor- or customer-reported, this is field data collection, not DER management, and the release says nothing about connectivity-model accuracy or how AI condition calls are human-reviewed before entering GIS or FEMA storm documentation. (Source: EIN Presswire / Noteworthy AI, October 15 — October 15 entry)
🏢 Utility Programs & Resource Plans
Wood Mackenzie reset the avoided-capacity benchmark at the center of every DSM business case — in both directions. In the U.S., WoodMac puts four-hour battery storage 65–75% cheaper than new open-cycle gas peakers for a 2026 commercial-operation date, depending on whether state carbon pricing applies, while projecting gas-turbine prices to reach $600/kW by end-2027 — a 195% jump since 2019 — amid OEM backlogs of 35–116 GW at GE Vernova, Siemens Energy and Mitsubishi, and a gas-capacity “supply deficit cycle through the late 2030s” driven by data-center load. Single-axis-tracker solar is the lowest-cost new build in 43 of 48 modeled markets, and storage LCOE is expected to spike after the ITC phases out in 2038 before falling roughly 10% by 2060.
Both halves matter and they pull opposite ways. Because demand response and energy efficiency are valued against the cost of the marginal peaking resource they displace, rising turbine costs and multi-year lead times push the avoided-capacity value of demand-side programs upward — this is the strongest version of the DSM case available right now. But falling storage costs simultaneously establish a competing, dispatchable benchmark that DR must be measured against, and PJM’s EIT approvals (80% storage) show procurement already moving that way. The practical instruction for an IRP filing: run explicit sensitivities on turbine price escalation, turbine lead times (5–7 years), state carbon pricing and the post-2038 storage-ITC cliff rather than anchoring avoided cost to a single peaker proxy. The caveat to carry: WoodMac reports LCOE, not full capacity or system value — storage’s and DR’s firm-capacity contributions still require separate accreditation before either substitutes one-for-one for a peaker. (Source: Utility Dive, October 8 — October 12 entry)
The controllable-load base keeps growing without a tax credit, and whether it becomes capacity depends entirely on enrollment. AHRI’s July 2026 shipment data show heat pumps at nearly half of all HVAC units shipped and up 17.6% year over year — the second-strongest July on record — while electric water heaters outsold gas units by roughly 52,000 even after heat-pump water heaters lost their federal tax credit at the end of 2025 (electric water-heater shipments were themselves down 3.9% YoY). Since July 2016, electric water heaters are up 23% and heat pumps roughly 77%. The Advanced Water Heating Initiative’s “State of the Market 2026” notes manufacturers are adding communication features and optimizing thermal-storage tanks so operators can shed operating periods — which is precisely the grid-interactive water heating that VPP and demand-response programs convert into dispatchable capacity. AWHI expects IECC codes and state mandates (all-electric building rules, low-GWP requirements in New York) to sustain growth.
The IRP arithmetic is a two-sided ledger and should be presented as one: every incremental electric water heater and heat pump raises winter peak — a cost in the load forecast — and expands the addressable device base for load shifting — a benefit in the avoided-cost case. The net value depends on whether the utility enrolls these devices into managed programs rather than letting them add uncontrolled peak. This is the same enrollment lever Brattle identified above, approached from the device side: the capability is shipping into homes whether or not anyone orchestrates it. The article quantifies shipments but not megawatts of flexible load, so treat it as a leading indicator of the controllable-load pipeline rather than a program-level number. (Source: Utility Dive, October 5 — October 13 entry)
☀️ Distributed & Community Solar
Massachusetts is now spending clean-energy compliance dollars on heating bills, and the same order opened a net-metering review. Governor Healey declared an energy emergency by executive order on October 5 (reported October 8), directing agencies to use alternative compliance payments — the fees suppliers pay for missing clean-energy targets — to supplement federal LIHEAP funding and create a one-time benefit for middle-income oil-heat households, after home heating oil averaged $6.12/gallon, up nearly 75% year over year. The order temporarily halves Alternative Portfolio Standard ratepayer costs (about $20M in relief, with the administration signaling an APS wind-down), shifts the SMART program’s “distributed solar charge” onto ACP funds and off ratepayer bills, and orders a review of statewide net-metering compensation while preserving the goal of doubling solar by 2035.
The signal is double-edged and both edges are consequential for DER practitioners. Affordability pressure is now strong enough that a state will tap the same ACP pools that fund efficiency, storage, microgrids and building electrification — which is a budget risk to any program drawing on compliance-fee revenue, and a reason to stress-test program funding continuity the same way this series recommends discounting California’s residential battery VPP capacity for DSGS/ELRP funding continuity. At the same time the order elevates low-income efficiency, heat-pump conversion and net-metering economics as the policy levers under review, and the net-metering review in particular could reset the behind-the-meter DER value proposition that underpins residential VPP and DR enrollment in the state. Massachusetts is the jurisdiction carrying a 3.5-GW demand-side resource target by 2035 under Executive Order No. 654 — roughly 13% of New England’s 26.1-GW peak — so a change to BTM compensation there is a change to the enrollment base that target depends on. The order offers no DSM program-budget figures, so the competition-for-funds effect is directional rather than quantified. (Source: Utility Dive, October 8 — October 12 entry)
🔬 EPRI Research Spotlight
No new EPRI publication fell inside this week’s window — the fourteenth consecutive week without one, and the pattern itself is now the finding. A targeted direct check turned up no EPRI research release dated within October 12–16; the surfaced EPRI material was prior-vintage work (the 2024 technical update on Grid Services Provided by Distributed Energy Resources, the DERMS Control Architecture for Grid Services in the Distribution and Bulk Power Systems product, the fast-and-flexible demand-response work, and the DER-in-resource-planning product) plus EPRI’s standing NERC Large Load Actions material. Report this as cadence, not omission: EPRI’s research releases arrive on a slower rhythm than a weekly digest, and the institute’s contribution continues to reach the trade record through citation and personnel rather than through publication.
This week supplied the confirming instance twice. EPRI research is cited inside the AWS interconnection op-ed as part of the case for AI-assisted planning, and EPRI’s large-load material underpins the NERC computational-load track this series has been tracking toward its December board adoption and December 31 FERC filing. Both are EPRI appearing as the substrate of other organizations’ arguments.
The practical consequence is unchanged and worth restating, because it is the methodological risk of the quarter: the computational-load standards will be written and filed before the measured operational data exists. The series’ standing observation holds — the only large-scale measured event remains the 3,800-MW July 22 trip in which data centers disconnected too early, and the 96-MW Aurora reference installation in Manassas, Virginia still targets late 2026 with no acceleration or slip detected. Meanwhile this week produced two genuinely new pieces of operational evidence from outside EPRI: PJM’s September review (two events, targeted DR called through the post-sunset net-load peak) and the October 13–16 emergency (DR called first, large-load backup not needed). The measurement record the standards need is being generated by grid operators in real time; the research institution’s cadence is simply slower than the regulatory calendar it informs.
Also worth noting against EPRI’s own prior contribution: Baum’s September argument that microgrids must move from emergency-only to everyday dispatch got an unintentional field test this week. Order No. 202-26-50 reached auxiliary, standby, directly connected and battery-storage resources — the scattered backup fleet — but reached them by emergency fiat, conferring no accreditation, no settlement record and no capacity-market status. That is the emergency-specification/capacity-evaluation mismatch Baum named, demonstrated at the scale of a 13-state footprint.
🚩 Utility-Sector Relevance Flags
⚑ FERC Moved the Curtailment-Order Fight Into Your State Commission — File There, Now
Topic: Large-Load Curtailment Priority / Federal-State Jurisdiction
Relevance: FERC’s Oct. 9 order (Docket ER26-3515) rejected IRAS and BYOC on jurisdictional grounds, keeping only the 50-MW definition and the registry, and Rosner’s concurrence left the curtailment order to states, after voluntary DR. Analysts expect state-driven IRAS-style regimes (Ohio, Pennsylvania named); Voltus is already planning on it. Pennsylvania’s Oct. 1 tentative order is the only finished template.
Action Signal: Engage — This is the highest-value, most time-bound action on the board. Get into the PA docket (M-2026-3064961) and your own state’s equivalent with drafted language: enrolled, compensated DR must be called before unsecured large load, and the shed order must be written in the tariff rather than left to emergency discretion. Confirm the Pennsylvania Bulletin publication date immediately — the 30-day comment clock plus 15-day reply runs from it, not from the Oct. 1 vote, and it is still unverified.
⚑ Voluntary Demand Response Is Now First in the Emergency Dispatch Order — and It Was Actually Called
Topic: DR Dispatch Priority / Operational Validation
Relevance: Rosner’s concurrence put curtailment “only… after voluntary demand response is used up.” Four days later PJM issued a Pre-Emergency Load Management Action (Oct. 13), recalled non-firm exports and all maintenance outage tickets, and reported it had not needed large-load backup generation. Removing IRAS leaves aggregated DR and BTM resources relatively stronger in the stack.
Action Signal: Implement — Put the dispatch-order language into your DR tariff and your emergency operating procedures, and verify that the sequence your control room actually follows matches it. Then audit the portfolio against it: revenue-grade telemetry, settlement-ready M&V, and a documented call-and-performance record. The priority is only worth what you can prove you delivered when called.
⚑ The Shoulder-Season Net-Load Peak Is Now Capacity-Defining — Re-Spec Your Dispatch Windows
Topic: DR Program Design / Capacity Accreditation
Relevance: Three events in seven weeks, all in the post-sunset ramp: PJM peaked ~152,800 MW on Sept. 1 (outages 15,502 MW / 8.47%), just under 127,000 MW on Sept. 17 (outages 37,916 MW / 20.7%), then declared an emergency on Oct. 13 with ~61,591 MW offline and a ~900-MW margin on Oct. 14. PJM’s own conclusion: maintenance “can’t simply be deferred indefinitely,” and demand is outpacing new generation.
Action Signal: Implement — Dispatch windows, event-hour assumptions and accreditation built around summer afternoons undercount resources available at 5–8 p.m. in September and October. Re-run your DR capacity value against shoulder-season and evening-ramp risk, not summer-peak coincidence, and widen contractual availability windows accordingly before the next program filing.
⚑ Brattle’s 4% Benchmark Converts Into a Portfolio Test You Can Run This Week
Topic: Residential VPP Sizing / DSM Portfolio Benchmarking
Relevance: A 4% system-peak reduction via residential VPPs — already achieved by the top 10% of utility DR programs — would take national residential VPP capability from ~10 GW to 35 GW and yield $4–7B/yr in household benefits, cutting five-year bill increases by more than half. The lever is OEM-led enrollment of existing devices, not new hardware.
Action Signal: Implement — 4% of peak is ~40 MW of residential flexible capacity per GW of system peak; at ~$66/kW-year that is on the order of $2.6M/year per GW of peak before energy, T&D-deferral and ancillary stacking. Compute your own figure and report the shortfall in MW. Cite with both caveats attached: aggregator-commissioned, and a uniform-attainment scenario, not a forecast. Keep it distinct from the Oct. 2 Brattle/Uplight finding (23 of 90 MW from orchestration alone), which remains the stronger line in a DERMS capital request because it carries no enrollment assumption.
⚑ DOE Named a 35-GW Idle Backup Fleet — The Emergency Order Is an Argument for a Standing Program
Topic: Behind-the-Meter Backup / VPP Enrollment
Relevance: Order No. 202-26-50 reached auxiliary, standby, directly connected and battery-storage resources, synchronized to the BPS or not, as a last resort before/during an EEA-3, with critical facilities excluded and environmental relief narrowly scoped. DOE estimates >35 GW of unused backup generation nationwide. The order explicitly does not make these capacity-market resources.
Action Signal: Engage — Take the federal framing and the customer list it implies into a standing enrollment offer: metered, accredited, compensated dispatch for the same assets, rather than compelled operation that builds no settlement record. Lead with the asymmetry — compelled dispatch pays nothing and accredits nothing. Note the real conflict the order surfaces for limited-fuel or limited-energy assets: grid support competes with site resilience, which is the trade a DERMS dispatch layer must arbitrate and a contract must price.
⚑ The Avoided-Capacity Benchmark Moved in Both Directions — Run the Sensitivities, Don’t Pick a Proxy
Topic: Avoided-Cost Methodology / IRP Sensitivities
Relevance: WoodMac: four-hour storage 65–75% cheaper than new OCGT peakers for a 2026 COD; gas turbines to $600/kW by end-2027 (+195% since 2019); OEM backlogs 35–116 GW; a gas “supply deficit cycle through the late 2030s.” Storage LCOE spikes after the 2038 ITC phase-out. Separately, ~80% of PJM’s first 2.1 GW of EIT approvals is battery storage, and Jefferies expects the FERC ruling to keep capacity prices higher for longer.
Action Signal: Implement — Stop anchoring avoided cost to a single peaker proxy. Run explicit sensitivities on turbine price escalation, turbine lead times (5–7 years), state carbon pricing and the post-2038 storage-ITC cliff, and show DR against both benchmarks — the rising peaker and the falling battery. State that LCOE is not capacity value: storage and DR firm contributions still need separate accreditation.
⚑ No Supply-Side Resource in PJM’s Fast Track Arrives Before 2029 — Price Flexibility as Capacity Available by a Date
Topic: Resource Adequacy Timing / Demand-Side Positioning
Relevance: PJM’s first EIT approvals (2.1 GW: Engie 860 MW Crawford PA + 800 MW Morrow OH, LS Power 455-MW Hunterstown uprate) are all due mid-2029; EIT takes only projects >250 MW, operational within three years, ≤10/year, and sunsets end-2027. SoftBank’s ~10 GW is phased through 2032. Both PJM answers to the 6,831-MW 2028/29 shortfall are stalled.
Action Signal: Implement — The 2027–2028 gap years have no supply-side answer in this queue. Price every demand-side option as cost per kW available in 2028, not per kW-year, and make the timing argument the headline of the filing rather than a footnote to the cost comparison.
⚑ Two Approved Gigawatt-Scale Michigan Contracts Carry No Flexibility Terms — That Is the Intervention
Topic: Large-Load Contract Design / Flexibility Obligations
Relevance: MPSC approved Oracle’s 1.4-GW Saline Township campus (expected to raise DTE load ~25%) and Google in Van Buren Township, both with customer-funded upgrades but no reported curtailment, flexibility or DR terms. DTE reports ~6 GW more in pipeline. Consumers Energy’s CEO claims ~$7.50/month residential benefit per GW (a utility claim). AG Nessel is challenging Saline as “fast-tracked.”
Action Signal: Engage — A large-load contract without a flexibility obligation is a forgone flexibility resource. Intervene for standard terms — defined curtailment depth, notice period and M&V — in every large-load contract and tariff proceeding, and pair the argument with Michigan’s PBR momentum (SB 768, rates +10.42% YoY vs 4.98% nationally): under performance-based regulation a deferred-CapEx flexibility program becomes a way to earn rather than forgone return.
⚑ BYOC Deliverability-in-Zone Would Reprice Locational Distributed Capacity — Watch the Refiling
Topic: Capacity Deliverability / Locational DER Value
Relevance: FERC’s two specific BYOC objections were a 10-year capacity-accreditation lock-in and allowing BYOC capacity to be delivered anywhere rather than in the constrained zone where the load sits. FERC also floated a voluntary non-firm transmission service — flexible loads accepting curtailment risk for faster interconnection and avoided firm-capacity cost. A narrower Section 205 refiling is expected.
Action Signal: Watch — If deliverability-in-zone survives into a refiling, it is a direct premium on feeder-aware, locationally sited DER and storage — exactly what a DERMS exists to capture, and an argument for locational granularity as a procurement requirement rather than a nice-to-have. Track the refiling and the non-firm-service concept together; they are the two channels through which load flexibility could get a new federal currency.
⚑ PJM Will Count Dormant DR Sites and Incremental Batteries as “New” Capacity — Check Your Portfolio Against It
Topic: Distributed-Capacity Eligibility / Bilateral Procurement
Relevance: On the suspended 6.8-GW backstop, FERC delayed the effective date five months and remanded the centralized auction, widening the bilateral window. PJM agreed that previously DR-enrolled sites that sat out the prior year count as “new”, and that incremental capacity such as new batteries at existing DR sites counts. Auction terms (exact locations + 15-year availability) fit DR poorly, per Aurora’s Julia Hoos. FERC has not confirmed that distributed capacity offsets an IRAS obligation — and IRAS is now rejected.
Action Signal: Implement — Inventory dormant enrolled sites and planned battery additions at existing DR locations against these definitions before PJM’s Oct. 29 revised filing settles them, and route near-term aggregator value through the bilateral phase rather than the interconnection track.
⚑ Cornell’s Curtailment Credit Market Is a Design Reference Worth Reading Before Your Next Non-Firm Tariff
Topic: Flexible-Load Market Design / Non-Firm Interconnection
Relevance: A Curtailment Credit Market layered on non-firm interconnection lets a constrained data center buy verified, physically deliverable reductions from factories, large C&I, VPPs or aggregators of batteries/solar/EVs/building controls, cleared only if feasible on the affected lines. Modeled on three grids (one a simplified New York): trading avoided ≥ one-third of the economic damage of forced cutbacks on an illustrative 100-MW reduction. Modeled, not piloted (Applied Energy, Sept. 22).
Action Signal: Watch — The commercial insight is the one to act on now even though the mechanism is unbuilt: it makes the large-load customer with a curtailment obligation the buyer of distributed flexibility, which is a far easier sale than an avoided-cost argument to a commission. Keep it beside FERC’s voluntary non-firm transmission service as the two live designs for pricing load flexibility, and note what a real program would need that this does not yet have: deliverability verification and settlement machinery.
⚑ Joint-Action Agencies Are Setting Public Power’s DERMS Data Prerequisites — and Interconnection-Study Throughput Gates Everything
Topic: DERMS/ADMS Prerequisites / GIS Data Quality / Queue Throughput
Relevance: FMPA ran one RFP covering 32 member utilities (serving 2.5M+ residents, ~11% of Florida) for vehicle-mounted AI collection feeding asset inventory and GIS development — City of Starke: ~2,300 poles in ~8 hours. Separately, >2,000 GW awaits interconnection review with waits at ~5 years and rising; PG&E unified 80+ data sources across 20 systems to prioritize 17,000+ upgrades, cutting costs 20% in a $100M pilot; Duke doubled annual project evaluations.
Action Signal: Watch — An accurate connectivity model is the precondition for any hosting-capacity or locational dispatch function, and for public power that specification is now being written at the agency level — so engage the joint-action agency, not the individual utility. Treat the AI-planning claims as unverified (a vendor op-ed, no products named, author’s own account of results), but take the dependency seriously: every month of queue delay is a month of deferral value you cannot book.
📌 Sources
October 12, 2026 Entry
– Utility Dive — DOE Presses PJM on Ratepayer Protections From Large Load Costs (October 8, 2026) (movement — the PJM backstop was reported September 18, September 25 and October 2; the DOE filing is new)
– Utility Dive — PJM Approves Fast-Track Interconnection for 2.1 GW From Engie, LS Power (October 9, 2026) (first appearance)
– Utility Dive — 4-Hour Storage Cheaper Than Gas Peakers Across Global Markets: WoodMac (October 8, 2026) (first appearance)
– Utility Dive — Massachusetts Shifts Clean Energy Funds to Help Heat Homes This Winter (October 8, 2026) (first appearance; the Healey order is dated October 5)
October 13, 2026 Entry
– Utility Dive — FERC Rejects Most of PJM’s Large Load Interconnection Plan (October 12, 2026) (first appearance; resolves the ER26-3515 verification item this series has carried since August)
– Mgrid.org — FERC Rejects PJM’s Curtail-First Plan for New Large Loads, and Keeps Only the 50 MW Definition and a Registry (October 10, 2026) (deduplicated — same October 9 order as the item above; carried for the docket mechanics and the 6,831-MW / 138,318-MW auction figures)
– Latitude Media — PJM’s ‘Unexpected Twist’ Is a Win for Distributed Capacity (October 1, 2026) (movement on the backstop reported October 2; new elements are the dormant-DR-site and incremental-battery eligibility concessions)
– Utility Dive — Heat Pumps, Electric Water Heaters Continue to Gain Ground: Reports (October 5, 2026) (first appearance; AHRI July 2026 shipment data)
October 14, 2026 Entry — all four sources cover the single PJM October emergency and are consolidated into one item above (deduplicated)
– T&D World — Largest US Grid Operator PJM Triggers ‘Maximum Gen Alert’ as DOE Issues Emergency Order to Maintain Reliability (October 13, 2026) (first appearance)
– PJM Inside Lines — Operations Update: PJM Extends Maximum Generation Alert to Oct. 14, Calls on Demand Response (October 13, 2026) (first appearance; primary operator source)
– Microgrid Media — PJM Emergency Power Order Allows Backup Resources (October 12, 2026) (first appearance; the DER/BTM mechanics of Order No. 202-26-50)
– U.S. Department of Energy — Energy Secretary Secures Mid-Atlantic Grid Ahead of Anticipated Stressed System Conditions (October 12, 2026) (first appearance; primary federal source for the >35-GW estimate)
October 15, 2026 Entry
– Utility Dive — Scaling Residential VPPs Could Cut Electric Bill Increases by Half: Brattle (October 14, 2026) (first appearance; Brattle report dated October 13, prepared for Everyday Electric — a different study from the Brattle/Uplight analysis reported October 2)
– Bridge Michigan — Fact Check: What’s True and What’s Not in Michigan’s Data Center Debate (October 14, 2026; updated October 15) (first appearance; first DTE and Consumers Energy load and bill figures in the series)
– PJM Inside Lines — PJM Reviews September Hot Weather Operations (October 12, 2026) (first appearance; Operating Committee review of the August 31 – September 4 and September 15–18 events)
– EIN Presswire / Noteworthy AI — Noteworthy AI Statewide Grid Intelligence Program Now Available to 32 Municipal Utilities (October 15, 2026) (first appearance; vendor release — claims are vendor- or customer-reported)
October 16, 2026 Entry
– Cornell Chronicle — Trading Power Reductions Could Reduce Data Center Strain on Energy Grid (October 16, 2026) (first appearance; underlying study in Applied Energy, September 22, 2026 — modeled, not piloted)
– Utility Dive — PJM Plans to Study SoftBank’s Ohio AI Project Under State Agreement Approach (October 14, 2026) (first appearance)
– Utility Dive — Utilities Can Use AI to Solve Grid Backlogs: AWS Energy Lead (October 15, 2026) (first appearance; opinion piece, results unverified, no products named)
– Latitude Media — FERC Rejected PJM’s Data Center Curtailment Plan. What’s Next for BYOC? (October 13, 2026) (deduplicated — same October 9 FERC order; carried only for the two specific BYOC objections, the voluntary non-firm transmission service concept, and the industry path-forward read)
Method: All five weekday entries (Monday October 12 through Friday October 16) are present in the research log; no days are missing. Items are reported on the day they entered the log, with their original publication dates shown. Several carry pre-week publication dates: the DOE FERC filing and WoodMac analysis (October 8), the Massachusetts executive order (October 5), the PJM EIT approvals (October 8–9), the AHRI shipment data (October 5), the Latitude backstop analysis (October 1) and the Applied Energy study behind the Cornell item (September 22). Within-log deduplication: the four October 14 sources cover a single event — the PJM October emergency and DOE Order No. 202-26-50 — and are consolidated into one item; the three sources on FERC’s October 9 large-load order (Utility Dive October 12, Mgrid.org October 10, Latitude Media October 13) are likewise consolidated, with the two analytical pieces carried only for detail the news report omits. Cross-week check against all prior digests: the PJM reliability-backstop procurement was reported September 18, September 25 and October 2 and appears here only as movement (the DOE statement of position, and the dormant-DR-site and incremental-battery eligibility concessions); Everyday Electric’s rebrand was reported October 9, so the Brattle study commissioned by it is a new finding rather than a repeat, and is distinct from the Brattle/Uplight analysis reported October 2; the DOE Section 202(c) pattern has been tracked since August, but Order No. 202-26-50 is a new order and the first in a shoulder-season month. Zero source URLs overlap last week’s Sources section. Two items were chased by targeted search and could not be confirmed, and are reported as open rather than resolved: PJM has published no deployed demand-response megawatts or performance results for the October 13–16 window, and the Pennsylvania Bulletin publication date that starts the 30-day comment clock on the PA PUC’s October 1 tentative order remains unverified. EPRI direct check: no publication dated inside the window — the fourteenth consecutive week — with EPRI appearing instead as a citation inside the AWS op-ed and as the substrate of NERC’s computational-load track.