Leadership1 distinct publisher2 min readUpdated
PJM's filing would make any 50MW-plus site the first load curtailed unless it brings newly built supply, verified on PJM's calendar. Five other grid operators owe FERC the same answer.
The Board Room · Leadership desk
Compiled by The Board RoomSomething wrong?How this is made
The procurement decision behind a June 2027 site has to close long before June 2027. PJM wants evidence of new supply on March 1, final verification on April 1, and only then service on June 1 [8]. Forbes columnist Robert Szczerba, who worked through the filing, says project teams fix on the service date and never notice the proof date sitting three months in front of it [9].
PJM will accept four kinds of proof [7]. Two of them are unavailable to anyone who starts shopping now. GE Vernova's turbine slots are booked through 2031, with its backlog at 116 gigawatts this summer [10], which is at least four years the wrong side of the proof deadline [2]. Connecting a new plant is slower again: more than five years for half of all projects [11], and of everything that entered those queues between 2000 and 2020, only 13 percent was running by the end of 2025 [12], meaning 87 percent was not [3]. Building on site skips the queue, which is most of the reason to do it [13].
That leaves the uprate, the restart, and the pledge to cut demand on request. The last one deserves a second read, because it is the same physical behaviour PJM would otherwise extract at a discount: qualifying curtailable load earns the full rate, while interim service earns half and PJM has floated waiving even that [6]. The identical willingness to go quiet is worth double depending on which side of the qualification line it was documented [4].
PJM's reason sits in its capacity auction, which hit the price ceiling and still came up 6,831 megawatts short, the second consecutive miss [14]. Measured against the filing's own threshold, that gap is about 137 sites of 50 megawatts [1]. The operator's argument is that interim service is a way to connect now rather than wait years for supply that does not exist, with cuts confined to real shortages [15]. The trade is that the customer carries the risk of being cut until it delivers the new megawatts on PJM's schedule [18].
Texas, which sits outside federal reach and did not wait to be told, froze its queue on August 3 and put 250 to 300 projects seeking 200 gigawatts under audit, more than double the most power the state has ever drawn at once [17]. PJM, which runs the grid for 67 million people [3], filed on August 13 [2], 56 days after the order and four days inside its window [5]. Whoever signs an AI capacity plan from here is signing a generation project with a documentation date attached.
Follow any of these and your For You feed starts watching them — no settings page required.
Ranked by verification strength, evidence, and original report placement.
On June 18 federal regulators ordered all six of America's regional grid operators to justify their rules for connecting very large customers, or rewrite them, giving them sixty days.
PJM filed its response on August 13, answering first among the six operators.
PJM has proposed that any load of 50 megawatts at a single site without its own power supply be placed on what it calls interim service, going first when the grid is tight, ahead of the factories and businesses PJM currently pays to power down.
PJM would cut only the megawatts it needs, only in the zone where the shortage is, and only while it lasts.
Everyone else who cuts back earns the full rate; interim service earns half, and PJM suggests the customer might like to waive even that.
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Specific but single-sourced
The factual spine is unusually specific — dated order and filing, a named threshold, a three-date calendar, and hard figures for the auction shortfall, turbine backlog and queue completion rate. But all of it arrives through one contributor column at one publisher, with no linked FERC docket, no quoted tariff language, and no second outlet or counterparty confirming any number. Nothing contradicts the account; nothing independently corroborates it either.
Regulatory action real, rule not yet in force
Concrete institutional steps have already happened: the FERC order was issued, PJM filed, and Texas independently froze its queue and launched a 200GW audit. That is genuine regulatory adoption of the 'conditional access' posture. But the interim-service tariff is a proposal awaiting approval, service does not begin until June 2027, and there is no evidence in the source of any data center, hyperscaler or developer having accepted interim service, qualified supply, or restructured a project around the March 1 date.
Headline outruns the article's own caveats
The headline and lede ('would go dark first', power 'the turbine market can't deliver in time') are more absolute than the body, which concedes curtailment is limited to the needed megawatts, the affected zone and the shortage duration, allows four qualifying supply routes including a demand-cut commitment, and calls PJM's rationale 'a good one'. The overstatement is modest and self-corrected within the piece, and the underlying constraints — 2031 turbine slots, five-year interconnection, a price-ceiling auction still 6,831MW short — are real, so the gap is positive but small.
Attention-seeking advisory framing, no disclosed stake
The column is written to be urgent and actionable — an alarming headline, a named deadline to 'own', and repeated self-citation of the author's earlier pieces on Big Tech's queue fear and Virginia's data center tax — which rewards escalation over caveat. Against that, no vendor, sponsor or financial interest is evident anywhere in the source, and the piece voluntarily relays PJM's counterargument and the limits on curtailment, which a purely promotional item would omit.
Low-moderate: one publisher, one document, unverified status
The chronology is internally consistent and arithmetically checks out (June 18 plus 56 days is August 13; 6,831MW is about 137 fifty-megawatt sites), which supports the reporting's care. Confidence is nonetheless capped by having a single publisher and a single article, no primary documents, no comment from any affected party, and no statement of where the filing stands in FERC's review — the one fact that determines whether the March 2027 deadline binds at all.
leadership
The same lies, every fire season: a coalition report makes the case for pre-planned crisis comms1 distinct publisher
invest
A stock that doubles on one readout was telling you what it thought of the odds1 distinct publisher
leadership
AI capex outgrew the consumer. Your demand forecast is now an AI bet.1 distinct publisher
leadership
The $100 sorority girl and the end of buying attention by the view1 distinct publisher
Distinct publishers with included, body-backed reporting in this cluster.
1 article · August 21, 2026