Signal Scanner · ENERGY, INFRASTRUCTURE & CLIMATE RESILIENCE · 11 August 2026

The Queue Was Never the Demand: Collateral Now Decides Which Data Centres Count

Interconnection queues are being audited and collateralised across the United States, and disclosed data-centre load is falling by a third or more wherever the test is applied. Grid planners, ratepayers and infrastructure investors are exposed to the gap between a request and a commitment.

The consensus is that American electricity demand has entered a step change, and that the grid's problem is building fast enough to serve it. The queues support that reading: ERCOT alone is tracking more than 474 gigawatts of connection requests. What a queue does not show is how much of that demand anyone has agreed to pay for. Over the past ten weeks, regulators and utilities have started asking, through deposits, transmission security agreements and state audits. Where the question has been put, the answer has come back smaller by a third to a half. Nothing about compute has been disproved. The number that capital plans are built on has moved.

Signal Identification

A regulatory pivot with immediate accounting consequences. What changed is not the demand estimate but the filter: the figure a utility discloses to its regulator and its investors now depends on whether a customer has posted money. Queue volume and contracted load have separated, and the gap is being closed downward, in public, quarter by quarter.

Time horizon: 1–4 years (PUCT decision 20 August 2026; RTO tariff filings late 2026; procurement consequences 2027-2029)
binds 1-2 yrs2026202720282030
Plausibility band: High
LowMediumHigh
Geographic / Jurisdictional Scope: Primary: the United States, sharpest in ERCOT, PJM, Ohio, Indiana, Georgia and New York. Spillover: any jurisdiction whose network planning treats connection requests as a demand forecast, including Ireland, the Netherlands and the United Kingdom.
PrimaryERCOT / TexasPJMOhio & IndianaGeorgiaNew York
SpilloverIrelandNetherlandsUK
Sectors exposed:
Regulated utilities and network planningData-centre developers and hyperscalersInfrastructure debt and equityTurbine, transformer and switchgear supplyState utility commissionsIndustrial site selection

What's Changing

The test is now measurable. Reviewing large-load tariffs in mid-July, RMI found that developers had requested 30 GW from AEP Ohio, nearly three times its typical peak demand of 10.5 GW, but signed only 5.6 GW of new contracts once the tariff applied (RMI, 17/07/2026). System-wide 2030 forecasts fell by roughly one-third at AEP Ohio and about half at Indiana Michigan Power. Over 80% of Georgia Power's 68 GW pipeline, more than 60 GW, sits in early technical review with no binding commitment.

Exelon put a price on the distinction. Its high-probability data-centre load fell nearly 40%, to about 11 GW, from 18 GW at the end of last year, once it began signing transmission security agreements carrying credit obligations and shortfall payments; about 4 GW of signed load has posted $1 billion in collateral (Utility Dive, 31/07/2026). The wider pipeline fell to about 25 GW from about 43 GW disclosed in May, inside one quarter.

Regulators followed. Texas ordered an audit of every data centre in a queue holding more than 474 gigawatts across 1,800 projects, about 90% of it data centres (The Texas Tribune, 03/08/2026), and ERCOT suspended the Batch Zero classifications due on 7 August 2026 (ERCOT, 03/08/2026). New York paused discretionary permits for hyperscale sites and asked its regulator to weigh an insurance pool against speculative large loads (Office of Governor Kathy Hochul, 14/07/2026). FERC gave all six RTOs 60 days to justify or rewrite their large-load tariffs (FERC, 18/06/2026).

What the test costs: requested load against committed load

Gigawatts before and after a commitment test Upper bar: as requested or previously disclosed. Lower bar: contracted or restated. AEP Ohio requests 30 GW New contracts signed 5.6 GW Exelon high probability 18 GW After security agreements 11 GW Exelon pipeline, May 43 GW Exelon pipeline, Q2 25 GW

AEP Ohio figures from RMI (17/07/2026); Exelon figures from Utility Dive (31/07/2026). Bars scaled to a common gigawatt axis.

Disruption Pathway

The pathway runs through disclosure, not construction. Stage one, complete in several jurisdictions, is instrumentation: a tariff, a security agreement or an audit that makes a developer choose between paying and withdrawing. Stage two, running through 2027, is restatement, as utilities revise the forecasts filed in rate cases and resource plans to the contracted number, and RTO tariff filings settle what counts as a commitment. Stage three, from about 2028, is reallocation: capital plans, turbine orders and network upgrades sized to the earlier figures are either defended on the smaller base or written down.

Stress concentrates in three places. Developers who filed across several territories now pay to hold each one; ESIG notes that coordination between utilities and grid operators is inadequate, producing duplicated effort and inconsistent visibility into cumulative system impacts, which is what made duplicate filing cheap (ESIG, 19/06/2026). Utilities that pre-committed capacity against uncommitted load carry the recovery risk into their next rate case. Equipment makers hold backlogs built partly on requests now being filtered. Two adaptations follow. Procurement moves from one forecast to tranches conditioned on signed service agreements, deposits, collateral postings and construction progress (RMI, 06/08/2026). And the cost of being wrong shifts toward the party that made the request.

Why This Matters Now

Boards and investment committees hold capital plans, site strategies and supplier contracts calibrated to numbers that three separate mechanisms are now revising downward. The input needing revision is demand itself. Most infrastructure cases in this sector still read an interconnection request, a letter of intent or a queue position as evidence of demand, and almost none carry a view on what share of the pipeline has money behind it. For a utility the exposure is a rate case in which the regulator asks why capacity was procured against uncommitted load. For a developer it is the cost of options that used to be free. For suppliers and lenders it is a backlog whose customers are being asked, for the first time, to show they intend to build.

Decision-action posture for this signal: Decide — the instruments are in force, four major jurisdictions moved inside ten weeks, and the Texas commission rules on ERCOT's Batch Zero timetable on 20 August 2026, so a capital plan still anchored on requested rather than contracted load is being decided by default.

Counter-Argument

The strongest objection is that this measures disclosure, not consumption. Nothing here says less electricity will be used; it says utilities have stopped counting requests they cannot bank. RMI reads the filtered forecasts as better evidenced rather than simply lower: ERCOT still carried 37% of requested load into its 2030 adjusted forecast, and AEP Ohio's contracted book stands at 17.8 GW through 2035 against a 10.5 GW historic peak (RMI, 17/07/2026). Texas ordered a verification, not a cancellation, and ERCOT's suspension awaits a commission decision (ERCOT, 03/08/2026). The same load may simply arrive later.

That is right about the physics and wrong about the capital. Investment decisions are made against the disclosed number, not the eventual one, and that number moved twice in one quarter at a single utility. A planner who books 43 GW and then books 25 GW has not learned that demand fell; they have learned the first figure was never a commitment. The projects that fail the test had capital plans, turbine slots and network upgrades standing behind them.

Implications

This is durable rather than transient, because the instruments driving it are contractual and legislated rather than sentiment-driven: tariffs approved by commissions, agreements filed at FERC, and a federal show-cause process running to a deadline (FERC, 18/06/2026). The inflection window runs from now, as second-half filings restate forecasts, to about 2028, when procurement decided on the old numbers either lands or is written down. Utilities that priced early gain, because a contracted book is easier to defend in a rate case than a queue position. Developers holding optionality across several territories lose it. Suppliers and lenders sitting behind uncommitted load carry a risk that has, until this quarter, been invisible in the disclosure.

Early Indicators to Monitor

Disconfirming Signals

Strategic Questions

Keywords

Large-load interconnection; speculative interconnection requests; transmission security agreements; data-centre load forecasts; ERCOT Batch Zero; FERC show-cause orders; large-load tariffs; collateral and deposits; stranded network assets; integrated resource planning; ratepayer cost allocation; grid capital plans

Bibliography

Source tiers: Tier 1, governments, regulators and intergovernmental bodies. Tier 2, think-tanks, academic institutes, major consultancies and quality data providers. Tier 3, quality journalism and specialist trade press. Tier 4, vendor, company and practitioner sources, used only as directional corroboration.

This Signal Scan is one component of a Shaping Tomorrow intelligence cycle on Energy, Infrastructure & Climate Resilience. Every monitored topic also includes monthly Change Trackers that assess what is materially shifting in momentum, and a quarterly Decision Intelligence briefing that converts the cycle into board-level strategic implications. Cycles may also close with Action Triggers, an end-of-cycle synthesis that reads across the three formats and surfaces patterns no single deliverable carries on its own. Continuous intelligence on this topic is available as a Single-Topic or Multi-Topic programme.

Discuss a Tailored Programme

Prepared by Shaping Tomorrow: 11 August 2026