Duke's Gigawatts Are Fine. Its Permission Slips Aren't.
Duke Energy's third quarter was, on its face, the most boring kind of good news. Adjusted EPS of $1.99 against $1.81 a year ago, 2026 guidance narrowed to $6.70–$6.80 with an explicit nod to the top end, the 5%–7% growth rate through 2030 reaffirmed, and the promise of the upper half of that range beginning in 2028 held intact. The $1.1 billion of net income, up 14% year over year, and an operating margin of 27.0%, up 2.6 points, say the machine is running.
But the interesting part of this call is not the quarter. It is that in the same seven days, Duke got told no twice — once softly, once not — on the two regulatory instruments that convert its once-in-a-generation load story into rate base. Demand is no longer the constraint. Permission is.
The pipeline is intact; the permission slip is not
Harry Sideris spent most of his prepared remarks on the industrial logic: a late-stage pipeline of 15.4 gigawatts to be converted into electric service agreements by the first half of 2027, 5.2 gigawatts of data centers already under construction and scheduled to begin taking energy in the second half of 2027, and $5 billion to $10 billion of upside capital waiting on signature in Indiana and Florida. That framing is a direct continuation of August, when he said “we're looking at landing all of that 15 gigawatts by the first half of next year” — Harry Sideris, President and CEO · 2026-08-04. Same number, same deadline, one quarter later. Consistency is good; a slipping deadline would not be.
What is new is friction at the state level. Shar Pourreza opened the Q&A on Florida, where the Public Service Commission this week declined to act on Duke's filed large load tariff and sent it back for revision.
We're making adjustments to our tariff and plan on refiling that. In the meantime, we're continuing to talk to our customers... So no slowdown in what we're doing in Florida. We'll have to refile the tariff, but we feel confident that we'll be able to reach agreement with the feedback that we received.
Then came Carly Davenport's question on the Smith combustion turbine, and the answer was blunter than the prepared remarks: “Our Smith CPCN that we filed was denied by the commission.” — Harry Sideris, President and CEO · 2026-11-05 Harry attributed it to the commission wanting the IRP to validate load first, called it small and low-cost, and pointed to the Anderson County approval as evidence the gas build still has support — plus newly prefiled CPCNs for roughly 2,800 megawatts at Davie-Davidson. Read together, though, the picture is of regulators in Duke's own backyard becoming the binding constraint on the very development projects that justify the capital plan. The Customer Protection Plus framework — minimum demand charges, termination penalties, credit support, capital advances — is the company's answer, and it is a good one. It just has to be accepted by someone other than Duke.
Financing a $103 billion build into a rate-hike tape
Here is where the macro bites. The global keyword screen for this quarter is dominated by the return of tightening — Fed rate hike sits near the top of the market-wide list, alongside "impact of high rates." That is the worst possible backdrop for a company whose non-current debt has climbed to $82.2 billion while its interest coverage has compressed to 2.1x.
The spend is not slowing. Quarterly capex of $4.2 billion, up 27% becomes, on a roughly $1.50 of capital outlay for every dollar of operating cash flow. Brian Savoy's answer is not to raise common equity at a discount. It is to sell minority stakes at premium multiples. Brookfield is accelerating $1 billion of its Florida investment from 2028 into this December, with $2 billion more in 2027: “bringing in $8.5 billion to delever Holdco when interest rates are moving against us is a great thing” — Brian Savoy, Executive Vice President and CFO · 2026-11-05 — and, crucially, we sold this business, the minority stake at near a 30 P/E. That is why holding company interest avoidance has become one of Duke's top-ranked themes this quarter while "financial risk" has dropped out of the top thirty entirely. The $10 billion of planned equity is a third addressed via $825 million of ATM pricing, $2 billion of mandatory convertible units in August, and a $125 million annual DRIP from 2027 — dilution deferred, not avoided.
Against that, the market backdrop is striking. The last ninety days have been brutal for the AI-power complex — AI data centers as a keyword cluster is down roughly 31%, data center revenue names down nearly 39% — while Duke's own tape reads +3.4% over the same window, a slow grind higher after touching 52-week lows in late summer. Duke has quietly become the anti-AI trade: contracted, tariff-protected, minimum-take revenue with a regulated return, rather than merchant exposure to a buildout whose equity holders are being repriced.
What went quiet
The loudest silence on this call is nuclear. In the second-quarter materials, new nuclear was the single highest-ranked Duke keyword, with momentum to match. This quarter it has vanished from the top thirty entirely. Sideris has not changed his position — and he said so explicitly, twice — but the script has now been performed on five consecutive calls: 300 megawatts of uprates, license extensions to 80 years, two of six plants approved, discussions with hyperscalers and Washington about risk-sharing. The conditionality is unchanged.
We're not going to go forward with anything until we have the protections for our customers and investors, until we have clarity in the supply chain and the executability of these projects, as well as being able to protect our balance sheet.
That is disciplined. It is also, at this point, a holding pattern rather than a catalyst. Where the emphasis has actually shifted is gas: natural gas is now Duke's number-one theme, with more than 5 gigawatts under construction in the Carolinas and Indiana, another 5 gigawatts in development, and a liquefied natural gas storage facility under evaluation at Davie-Davidson to blunt price volatility. The speed to power mantra of prior quarters has given way to something slower and more procedural.
One more tell. Shar Pourreza asked in August whether Duke would consider a Genco structure in Indiana to bypass CPCNs; Harry said “we are looking at that” — Harry Sideris, President and CEO · 2026-08-04. Yesterday the same analyst asked the same question, and the answer was “we're open to that” — Harry Sideris, President and CEO · 2026-11-05 — marginally warmer, still noncommittal. When a question gets asked identically two quarters running, it is a live pressure point, not a hypothetical.
The bottom line
Brian Savoy told investors in February that Duke is “risk-adjusting our load forecast basically to the minimum billing demands of these contracts” — Brian Savoy, Executive Vice President and CFO · 2026-02-10, and that remains the core of the equity case: the growth is contracted, the counterparties are hyperscalers, and the 2028 inflection is underwritten by minimum takes rather than hopes. The Q3 beat and the top-end guidance narrowing confirm it. But the delta between Duke and a pure data-center derivative is now entirely regulatory, and regulators just moved twice in a week. Watch three things: the North Carolina orders expected by mid-November on both rate cases and the large load tariff; the Florida refiling; and the February roll-forward, where the $5–$10 billion of upside capital, the Indiana multiyear rate plan, and any nuclear decision all land at once. Meanwhile, $0.22 of year-to-date weather benefit is being partly reinvested to derisk future years — a modest but deliberate sandbag, and the kind of thing you do when you intend to be believed.