Reference

Latest Updates#

This section logs every change to the report since publication. Entries are newest first. Every item is dated, carries the same confidence tag used throughout, and names the sections it revises — and each one has already been worked into the body at the point of use, so the report reads as a single current document rather than a base text plus errata. This page indexes the changes rather than housing them.

Revision history#

Table L1 — Revision history. One row per revision, with the sections touched. The blocks below expand each entry; the body already carries every change.

Sources: FERC; NERC; ERCOT; PJM; DOE, as listed in the Catalog of Orders, Rules, and Directives Cited.

v1.2 — developments of July 20–21, 2026#

The July 20 resource-adequacy reports came due, and two of the six were recovered in time for this revision — one from the filing itself, one from reporting of it. Taken together they are the first evidence of what FERC's regional route produces in practice, and the two answers point in different directions. CAISO reports no systemic generation-adequacy shortfall in California and attributes it to an integrated forecasting, procurement and planning framework; §2 carries that as a comment rather than a finding, since the region carries a fraction of the demand pressure and its procurement mandate is a feature of California's regulatory structure rather than a tariff term available to an RTO. MISO answers the same question in the opposite direction, and consistently with its position in the record: an expedited study process for large loads run outside the interconnection queue as the ERAS fast lane winds down, a non-firm transmission service option, and a published target of approval within 120 days once studies are complete. The region reporting no adequacy gap proposes least; the region with the fastest data-center growth proposes most. §2 now carries both. Two further developments are procedural. CAISO's published calendar ends in a November 16 filing and states that it assumes an abeyance, so two of the six regions now look likely to answer in mid-November rather than on August 17. And PJM's backstop procurement acquired dates — a window of September 10 to October 9, 2026 in the June 30 design paper, with cost allocated to the load additions that created the need — while the 2028/29 shortfall is now stated precisely at 6,831 MW against a 14.7% reserve margin. Neither item reverses a conclusion of the report. One narrows the geography of a finding; one qualifies the uniformity of the federal deadline, in the direction Section 7 already argued.

Table L2 — Developments of July 20–21, 2026, and what each does to the report.

Sources: FERC show-cause orders of June 18, 2026 (EL26-67 through EL26-72); ISO New England compliance notice, June 29, 2026.

v1.21 — the reach of the voluntary commitments#

One addition to Section 3, prompted by a question the report had not answered directly: what the hyperscalers' own cost commitments actually reach. The Ratepayer Protection Pledge binds seven consuming companies, not the colocation developers who own much of the capacity in the queue and not the utilities that divide the costs — and even a willing payer is billed by a tariff that, under the Commission's long-standing transmission pricing policy, may recover the upgrade from everyone. Neither point changes a conclusion. Both sharpen why Section 3 treats the pro forma cost-recovery agreement, rather than any voluntary undertaking, as the deciding instrument.

Sources: The White House; Google; Harvard Electricity Law Initiative; Brookings.

v1.22 — editorial pass and a new annex section#

An editorial pass on reviewer comments, with one structural addition. The account of CAISO's July 20 filing was cut to a comment and restated: it reports no systemic adequacy shortfall and attributes that to California's integrated forecasting, procurement and planning framework, but it does not argue that no further reform is needed, and it commits the additional work to its own stakeholder initiative — a distinction the previous wording blurred. MISO's filing now leads the passage, which is the right order of importance. A duplicated PJM backstop entry on the near-term watch list was consolidated and two rows put back into date order. The confidence key gains a fourth tag, Speculative, for genuine forecasting: Emerging had been carrying proposals, forecasts and opinions at once, and the long-horizon watch-list items have been retagged accordingly. Figure D2's caption now ends on a bolded Finding line. And the technical annex gains T.11 on operational impacts — situational awareness and the EMS, ramping and dispatch, regulation and AGC, reserve deployment and the load-side contingency — the one subject area the report described in planning terms without ever reaching the control room. No conclusion changes.

v1.27 — the two arrangements behind Section 5#

Section 5 set out the co-location rules without naming the transactions that produced them. This revision adds that record from the primary documents. Three findings change how the section reads. The Susquehanna amended ISA was rejected for tracking PJM's generally applicable guidance rather than for any substantive defect, and expressly without prejudice, so the December 2025 order answers a question the Commission had declined to reach case by case. Both nuclear arrangements large enough to test co-location — Talen and Amazon at Susquehanna, Constellation and Microsoft at Crane — now take front-of-the-meter network service, and both settled on that structure before the replacement rate existed. And the Capacity Interconnection Rights adjustment that the December order treats as a write-down operates elsewhere as a transferable asset: the June 1, 2026 waiver moved 760 MW from Eddystone to Crane and advanced the restart by about three years. A November 2023 fallback event at Susquehanna, in which the co-located load drew from the grid for several hours during a unit outage, is added to the discussion of host unavailability as the only operating record the debate has.

Three further corrections followed on review. In Section 4, the December 2022 west Texas event now sits in a note of its own. It is the largest in the record and the least representative of it: the fault escalated to a three-phase fault caused by a breaker failure, so the clearing was not normal, and the load was mixed, with substantial losses from oil and gas facilities whose drives trip on undervoltage for reasons unrelated to the ITIC curve. The name compounded the difficulty, since NERC published its Odessa Disturbance reports about solar inverters rather than load; this report now calls the load event by its region. The July 2024 northern Virginia event is stated as the archetype the ride-through rules address. In Section 6, the Duke curtailment rates are restated on the basis the study actually uses — shares of annual energy, not of running time — which puts the hours in which some curtailment falls at 85, 177 and 366 rather than 22, 44 and 88, mostly as partial reductions, and the headroom at 76, 98 and 126 GW. The study's own first-listed limitation, that it does not model network constraints, is now recorded wherever the figure appears. And a new subsection distinguishes curtailment an operator calls from deferral the load schedules for itself, which is the trade-off Finding 5 now carries.

v1.1 — developments of July 13–17, 2026#

Four actions landed in five days, and all four point the same way the report already did: the reliability track (registration plus standards) now carries a federal directive and a hard deadline, while the political-economy track is hardening faster and less predictably than the tariff track. Nothing in this update reverses a conclusion of the original report. One item changes a confidence tag; one corrects a factual detail; two reinforce existing assessments.

Table L3 — Developments of July 13–17, 2026, and what each does to the report.

Sources: FERC; NERC; ERCOT; PJM; DOE.

What the update changes, in one paragraph#

RD26-7-000 matters most, because it acts directly on a central reform in the current reliability program. The Synthesis identifies NERC registration, rather than any tariff, as the discipline that reaches a large load wherever it sits, including when it islands off-grid (§4, §8). Until July 16 that reform rested on NERC's own accelerated plan and was tagged Likely. It is now a federal directive with two dates attached, which moves the existence and timing of the obligation to Settled while leaving the registry threshold and the standard's substance genuinely open. The second PJM shortfall, meanwhile, is the first hard market evidence that the 2028–2031 stress window in §9 is materialising rather than remaining a forecast — and the New York moratorium, arriving by executive order in the same week as a broadened federal pledge, is §7's community-consent thesis playing out in real time.

Filing the next update#

Revisions within this publication cycle are numbered v1.21, v1.22 and so on, so that the second decimal marks an update to the current edition rather than a new one. Each adds one row to the revision-history table and one short block below it, using the same four fields: Development (the instrument, with docket or citation), Date, Effect on this report (which conclusion it strengthens, weakens, or corrects, stated plainly — including “no change”), and a confidence tag. The body edit should be made at the same time and marked in the revision-history row, so that this page always describes changes that have actually been integrated. Where an update contradicts a prior assessment, the original judgment should be left visible and dated rather than quietly overwritten — the value of a report in a fast-moving docket is partly the record of what looked true when.

Data in this version is current through July 21, 2026. The July 20 resource-adequacy reports and the July 24 ERCOT submissions have now come due; the August 3 abeyance requests and the August 17 show-cause responses fall within the next month, and either may change the picture materially.