One site in Texas, three states on our wire
Start with the confession, because it’s instructive. Eleven of this fortnight’s 28 items are one event: on October 9 the NRC published a draft environmental impact statement for Fermi America’s Project Matador, four Westinghouse AP1000s on the Advanced Energy and Intelligence Campus in Carson County, Texas, next to DOE’s Pantex Plant. Our enrichment placed that campus in Texas, Oklahoma and New Mexico, sized it at 4.0, 4.5 and 4.7 GW, and in one entry called the AP1000 an SMR. It’s none of the last three things.
The error has a cause, which makes it more interesting than embarrassing. The draft EIS comes with Section 106 consultation letters to the Comanche, Kiowa, Tonkawa, Mescalero Apache and Jicarilla Apache, among others, and each letter carries its tribe’s address. The model read the envelope and filed the reactor where the mail was going. That’s the sixth straight issue with a geography error on our wire. It’s also the first one where the fix is obvious: take the state from the docket, not the correspondence.
The substance is real. The NRC’s preliminary recommendation is to issue the combined license, pending the safety review. Comments run through November 23, and the agency is aiming for a final EIS in December. Fermi wrote the draft itself under NRC supervision, as part of a pilot that lets applicants prepare their own environmental document. The application went in during the summer of 2025. Going from a submitted COL to a favorable draft EIS on four large reactors in about sixteen months is, by NRC standards, a sprint.
A license for 4.4 gigawatts, a lease for 222 megawatts
The nuclear piece of Matador is about 4.4 GW. The campus plan was 11 GW of onsite generation and, per later updates, closer to 17. The demand side looks like this: in August Fermi signed its first binding lease, with AI cloud provider TensorWave, for a data center supported by 222 MW, with room to grow to 650 MW. Over the same stretch, the company’s cofounders (CEO and CFO) left, and SEC filings say the CEO was terminated for cause. (Lease and leadership details are outside Dynawatt’s usual source set; they come from ANS Nuclear Newswire.)
None of that makes Matador unserious. It makes it typical. The licensing machine is now faster than the customer machine. This is the third issue running where the paperwork side of the ledger is long and specific and the demand side is one line. This fortnight’s one line is a 1,435 MW large-load request entering the MISO queue (J4331), study-eligible, customer unnamed. That’s the second MISO large load of that size since August’s 2,015 MW entry. A 300 MW request (J4320) came in alongside it with no fuel type or load flag, so it could be either.
The fast lane has a tipped fuel assembly in it
Three weeks ago we titled an issue “Restarts are the fast lane.” Palisades has since offered a footnote. Fuel loading started over the weekend of August 30, and during loading one of the 204 assemblies tipped. Loading is paused. The assembly is still in the vessel, Holtec is building retrieval equipment, and the plant is preparing a license amendment request to cover the retrieval method. The NRC says there was no radiological release and no one was hurt, and Holtec says the fuel is intact. Neither has given a restart date. (Outside our usual source set: WWMT and The National, both September 23.)
What showed up on our own wire was the NRC staff’s initial assessment, dated October 9, of 2.206 petitions filed against the restart between May 2025 and April 2026. That’s a procedural item, and it usually ends without action. But the timing captures the restart thesis in one picture. A restart is the fastest firm megawatt available right up until one component does something unexpected, and then it runs at the speed of a retrieval tool that doesn’t exist yet. The thesis still holds: Crane and Palisades remain years ahead of anything greenfield. It just turns out the fast lane has potholes, and they’re the same size as everyone else’s.
At the same address, Holtec completed the second part of its construction permit application for Pioneer 1 and 2 (two SMR-300s, about 300 MW each) on October 2. That’s the fourth procedural step on that docket we’ve covered since July. The new reactors’ paperwork keeps moving on schedule while the old reactor next door waits on a fuel assembly.
Meanwhile, the stuff that gets built
The largest single generation entry this fortnight is a 3,000 MW combined-cycle gas plant, Brazos Energy Security, in Fisher County, Texas (ERCOT 28INR0456). Nobody is writing an EIS pilot program for it, because nobody needs to. Next to it in the ERCOT queue sits a 1,320 MW steam stage for Blue Energy’s Victoria County project (33INR0007), which pairs GE Vernova gas turbines with planned BWRX-300s. Blue Energy filed the first part of its construction permit application on September 16, and the reactors are targeted for 2032. Gas first, nuclear later, sharing a turbine hall: a concept that admits out loud what the rest of the market keeps implying.
Round it out with a 500 MW battery entering MISO in Mower County, Minnesota (J4078), and the usual run of NRC housekeeping: Kairos asking for operator-licensing exemptions for its salt-cooled design, an X-energy emergency-planning-zone methodology revision, Westinghouse human-factors reports for the AP300, and two Oklo licensing filings. All necessary, none of it power. DOE marked the fortnight with a fact sheet announcing that a golden era of American nuclear energy has arrived. The queue suggests it’s arriving with a gas turbine in front.
PJM’s winter problem is a measurement problem
The most consequential item may be the least glamorous. At PJM’s Market Implementation Committee, Voltus raised a problem with the 2025 Annual Demand Response change (ER25-1525), effective for delivery year 2027/2028. The rule measures a demand-response site’s winter peak load at a fixed 9 a.m. hour. On winter critical-peak days PJM tends to dispatch demand response right around then, so a site that does what it’s asked records an artificially low baseline, and its countable curtailment capacity shrinks. Summer calculations add curtailed load back. Winter doesn’t. Fixing it needs a rule change and FERC approval before the 2027/2028 registration season.
It’s the old line about measuring a thing changing it, except here it’s PJM’s own rule doing the changing. And it lands exactly where PJM’s reliability risk has moved: from summer afternoons to winter mornings, the hours when data-center load stays flat and everything else peaks.
What to watch
Matador’s comment docket. Comments close November 23 and the final EIS is targeted for December. If the pilot holds that schedule, applicant-prepared EISs become the default, and every large-load campus with a reactor attached will want one.
Palisades’ license amendment. The retrieval plan has to clear the NRC before fuel loading can resume. The filing date is the first real signal of how long this pause runs.
Fermi’s second lease. 222 MW against 4.4 GW of nuclear and up to 17 GW of planned generation. The next binding tenant says more about Matador than the license will.
PJM’s winter peak load fix. It has to get through the stakeholder process and FERC before 2027/2028 registration opens. If it slips, PJM’s winter demand-response numbers will undercount by design.
Sources: NRC ADAMS and the Federal Register (Fermi America Project Matador draft EIS, 91 FR 64702, Docket NRC-2026-0100; Palisades 2.206 initial assessment; Holtec Pioneer construction permit filing; Kairos, X-energy, Westinghouse and Oklo submittals); MISO and ERCOT interconnection queues; PJM Market Implementation Committee materials. Outside Dynawatt’s usual source set: ANS Nuclear Newswire (Oct 9) for Fermi America leasing and leadership and Blue Energy; WWMT and The National (Sep 23) for the Palisades fuel-loading incident; World Nuclear News (Oct 2) for the Pioneer filing.
Dynawatt is an independent brief compiled from public sources: ISO/RTO interconnection queues, FERC notices, the ERCOT Large Load Working Group, PJM stakeholder committees, NRC ADAMS, and DOE Loan Programs Office announcements. It is informational, non-advocacy, and not investment advice.