Duke Energy's Nuclear Tightrope: Growth Is Confirmed — the Risk Question Is Not

The quarter: reaffirmed, and then some

Duke stepped into Q2 2026 with adjusted EPS of $1.43, up from $1.25 a year ago, leaning on large-load growth and heavy infrastructure investment across its regulated territories. Management kept the full-year band of $6.55–$6.80, the 5–7% long-run growth rate, and pushed the step-up story forward: “we are more confident than ever that we will deliver in the top half of the range beginning in 2028 when we expect to see accelerated growth from the economic development projects we have secured under ESAs.” — Harry Sideris, President and CEO · 2026-08-04 The bridge is visible in the numbers: Q1 revenue rose 11% year over year to $9.2B, and capital expenditure reached $4.1B, up 30% — the industry's largest regulated build at over $1B a month.

From megawatts signed to megawatts built: nuclear takes center stage

The most notable shift on this call was a re-weighting of the strategic agenda. Through 2025 and early 2026, the headline was ESA signings; Duke has now secured “7.8 gigawatts of electric service agreements with data center customers” — Brian Savoy, Executive Vice President and CFO · 2026-08-04, with the remainder of the 15.4 GW late-stage pipeline expected to convert to ESAs by the first half of 2027. The build is visibly underway — 5 GW of gas under construction, 2.5 GW more advancing, a GE Vernova turbine framework expanded to 26 units with the first delivered to Person County in July, and a plan now adding 15 GW of capacity by 2031. On this call, though, the data-center story was treated almost as a given — the real focus was new nuclear. The keyword spiked to the top of Duke's own momentum table (momentum 219, up from ~109 when it last surfaced in Q4 2024), bundled with financial risk at #2.

Harry Sideris was explicit that nothing happens until someone shares the construction risk:

We want to continue to emphasize that additional financial protections are needed before we would propose a new nuclear project. Any structure to advance new nuclear must address first-of-a-kind and supply chain risks, provide financial risk protections for our customers and our investors and ensure a strong balance sheet during the construction cycle.

The nuance: the AP1000, not the SMR, "seems to be in the lead" for Duke, even as it keeps an SMR early-site application at Belews Creek, a COL license for two AP1000s at Lee, and observation rights on OPG's Toronto project. This is a step-function from the prior posture — in Q1 management listed the same questions but stayed at arm's length. Now there is a live DOE engagement on supply-chain derisking and an acknowledgment that a commitment path is being evaluated. The Carolinas resource plan — with the load forecast now at the high scenario after the new ESAs — is the forcing function, with a commission order expected by year-end.

The affordability-defense toolkit grows

The second new thing is a careful expansion of customer-protection and capital-cost levers in the face of affordability rhetoric in Indiana and North Carolina: the "Customer Protection Plus" commitment (plus the Ratepayer Protection Pledge signed with hyperscalers in July), a May application for Department of Energy loans that could save "billions," and an accelerated one-year flowback of tax credits on a Florida battery project to offset a 2027 base-rate increase. The DEC settlement in North Carolina — 9.8% ROE, 53% equity, an earnings-sharing mechanism up to 10.3% — extends the multiyear rate plan framework. The playbook is familiar, but the environment is not; the February framing — “In the past, we have always tried to settle or settle portions and have a constructive track record of doing that. We will look to do that again” — Harry Sideris, President and CEO · 2026-02-10 — now has to work harder.

Worth watching: Harry reopened a genco structure for Indiana, something Duke has repeatedly waved off. “On the genco side, we are looking at that... that may be something that we're going to revisit in the future to be able to provide financing as well as another layer of protection for our customers.” — Harry Sideris, President and CEO · 2026-08-04 For a company that has answered "we don't need a genco," that is a genuine door-opening, and it dovetails with the financing machine: $600M priced under the ATM at "attractive pricing," DEF minority-interest proceeds, and an FFO to debt target of 14.5% heading to 15%.

Durability vs. the coverage constraint

The tension in the model is cash flow. Brian Savoy framed the payoff precisely: “the cash generation continues to grow and it's durable well into the late '30s” — Brian Savoy, Executive Vice President and CFO · 2026-08-04, with nuclear PTCs hitting parity around 2028-29 and investment returns then taking over from tax credits. But the near term is lumpy — operating cash flow fell 31% year over year in Q1. That swing is exactly why interest coverage matters: at 2.8x it sits well below its decade-ago ~5x level, and with $80.5B of long-term debt on the book, the "strong balance sheet during the construction cycle" clause in the nuclear framing is doing real work.

Why it matters

The backdrop sharpens the contrast. The global conversation this quarter is dominated by ERCOT interconnection queues (Batch Zero) and IEEPA tariff refunds — themes defining the merchant and data-center complex, and data center demand more broadly. Duke is nearly insulated from both: no ERCOT exposure, no tariff sensitivity for a regulated utility. Its growth is contractual, regulated, and visible in minimum-take ESAs. Yet the stock has drifted down ~9% off its March high on a quiet 90-day tape — the market is waiting for the load to actually ramp (2027-28) and, more pointedly, for the nuclear risk-sharing framework to resolve. The data-center supercycle is priced as a growth story; what isn't priced is who carries the construction risk of 15 GW of new dispatchable capacity — and a potential nuclear build — over the next decade. That is the open question Duke put back on the table, and why this call, more than a routine EPS reaffirmation, is one to read twice.