PJM says Oklo's fix left the 750-MW model unstable, asks FERC to deny

PJM Interconnection asked the Federal Energy Regulatory Commission on September 4, 2026 to deny Oklo's complaint over the removal of its 750-MW project from the grid operator's first reformed interconnection cycle, according to PJM's answer in docket EL26-101-000. PJM says the model Oklo submitted to cure a May 15, 2026 deficiency notice was unstable when PJM tested the whole facility at its 750 MW maximum output, that its tariff allows a single round of corrections, and that putting the project back into Cycle 01 could delay the cycle's Phase I study, scheduled to begin September 28, 2026, by as much as two months.

Oklo's emergency complaint, covered by The Duck Curve on August 31, argued that PJM flagged new deficiencies too late to be cured and applied a low-voltage ride-through test written for inverter-based resources to a facility that pairs 150 MW of advanced nuclear and 300 MW of natural gas generation with 300 MW of fuel cells. PJM's answer disputes both points. It reached the public docket on September 8, four days after filing; as of September 12 the docket sheet lists no Commission order and no reply from Oklo, plus interventions by PJM's Independent Market Monitor and Public Citizen.

The project's identity and site are redacted in the public filings; its point of interconnection is not. Utility Dive's August 31 report, noting the complaint did not identify the project, pointed to Oklo's planned 1.2-GW campus in Pike County, Ohio, for Meta. The public attachments to Oklo's complaint, re-filed on September 1 with privilege markings and what Oklo's cover letter calls no substantive changes, point elsewhere: PJM's May 15 deficiency notice, Attachment E, names Possum Point 500 kV and Ladysmith 500 kV as the two substations at the point of interconnection, on a line PJM gives as 47.56 miles long, and a kickoff-call email in Attachment I is captioned "PJM / Dominion / Oklo Power LLC."

Nuclear News, the American Nuclear Society's publication, reported on September 2 that both substations are just north of Richmond, Virginia, and quoted Oklo's head of media and communications, Bonita Chester: "The complaint relates to an early-stage potential generation project in Virginia." Chester also told the publication the filing "should not be understood as a deployment announcement, final investment decision, or commitment to a particular generation configuration."

What PJM says happened to the model

The answer quotes the May 15 notice as telling Oklo that "When initializing the full project facility with all generators turned on ... there are a lot of suspect initial conditions," and says the initial model showed a stable response at full output but with DSTATE errors that Oklo was asked to correct (a footnote defines DSTATEs as hidden internal values such as exciter states and rotor angles). The notice also cited transformer winding configurations, a missing single-line diagram, and mismatches in NextGen, PJM's application portal, with a June 1 response deadline.

What came back, in PJM's telling, made things worse. The answer says Oklo removed the under-excitation and over-excitation limiter models from all of the synchronous units, the nuclear and gas generators, rather than adjusting their parameters, and that "It appears Oklo tested the 'cured' model only with the inverter-based resource ('IBR') component (fuel cell) in service (300 MW at the POI)." When PJM ran the model with the entire project in service at 750 MW, it says, the nuclear and gas units showed unstable performance. PJM hedges the cause, writing that deleting the limiters "may have caused" the instability, and says the cure package omitted the Dynamic Model Report and checklist the guidelines require.

On Oklo's central legal argument, PJM says it "did not apply the low voltage ride through test to the facility." An August 7 email from PJM's Donnie Bielak, Attachment K to the complaint, had put it differently: the project was withdrawn because Oklo was "not able to demonstrate stability for all fuel types through the low voltage ride through (LVRT) test per the DMDG Guideline." The three-phase fault PJM ran is, per the answer, a standardized stress test; the project was rejected "because Oklo did not provide a working model for the full facility's MFO, consistent with DMDG, sections 7.1 and 7.3." In PJM's Dynamic Model Development Guidelines, section 7.1, "MFO Assessment," tells developers to verify that the requested maximum facility output is achievable at the point of interconnection, and section 7.3, "Flat Start Test," requires the models for "any type of generator" to be usable in the Siemens PSS/E software and to "flat start" through a 20-second no-disturbance simulation. Section 7.4, which Oklo relied on, is titled "Voltage Ride Through (VRT) Test (for IBRs)" and applies a three-phase fault for 9 cycles; PJM's answer calls the test it ran on the Oklo model its "standard 5 cycle, three-phase fault test."

One cure round, four other projects, and a two-month clock

On timing, PJM says the June 24 NextGen comments recorded inconsistencies between the revised model and the portal data on transformer windings and fuel-cell loads and reactive capability, problems that arose in the cure package and so could not have been flagged on May 15. The tariff's deficiency review, in PJM's reading, provides one notice and one response, and the comments in the August 3 withdrawal email were "advisory in nature." The emails Oklo cited as unanswered did not ask about the PSS/E model, PJM says: "If Oklo did not understand PJM's comments concerning the model, it did not communicate that to PJM."

PJM's answer to the discrimination claim is that Oklo cleared the bar four times. Four other Oklo applications in Cycle 01, PJM project numbers C01-1730, C01-1734, C01-1738, and C01-2157, had deficiencies, resolved them, and advanced to Phase I, totaling 900 MW of maximum facility output, PJM says, citing its status page as of September 3. PJM says 811 requests came in and 88% advanced through application review; its August 3 announcement put the accepted count at 715.

The Phase I base case was posted on August 28, and Phase I is to start after the 30-day review, on September 28. Reinstating Oklo would require a stand-alone deficiency review, a rebuilt cycle model, and a new 30-day posting, a delay PJM puts at "as much as two months" and says would ripple into Cycle 02. If the Commission grants relief anyway, PJM asks it to act by September 14, 2026 and grant the needed tariff waivers. The withdrawal was without prejudice, the answer notes; Oklo may reapply in a future cycle or, if it qualifies, seek PJM's Expedited Interconnection Track.

Outlook

PJM's requested action date of September 14 falls 14 days before the Phase I start; the docket showed no order as of September 12. If September 28 passes without an order, the answer's argument that later reinstatement "would exacerbate the cascading delays" only strengthens, which points toward denial or a remedy outside Cycle 01, such as the Expedited Interconnection Track PJM itself raised. The answer reframes the dispute: Oklo asked which stability test applies to a mixed plant, while PJM asks whether any applicant gets a second cure round, a question that reaches the roughly 12% of Cycle 01 applications implied by PJM's 88% figure.