North Carolina’s Republican-controlled Utilities Commission has drawn a rare line in the sand.

September 18, 2026, the North Carolina Utilities Commission denied Duke Energy Progress a Certificate of Public Convenience and Necessity for a new natural-gas simple-cycle combustion turbine at the Sherwood H. Smith Jr. Energy Complex in Richmond County. The denial is without prejudice, meaning Duke can refile. The message, however, is clear: data-center load forecasts and voluntary pledges are not enough if the utility cannot show, on this record, that everyday customers will not underwrite the plant.

The proposed unit is a roughly 240- to 255-megawatt hydrogen-capable peaker, priced at about $584 million, with a potential in-service date of January 1, 2030. It would have been the sixth combustion turbine at a site that already hosts five simple-cycle units and two combined-cycle units totaling more than 2,200 MW. The plant sits across from Amazon’s planned $10 billion, 21-building data-center campus near Hamlet, southeast of Charlotte. Duke did not formally assign the turbine to Amazon, but commissioners treated the linkage as obvious: much of the utility’s projected load growth is data-center driven.

Why the Commission Said No

The 3-1 decision (Commissioner John Gajda did not participate; Commissioner Floyd McKissick dissented) turned on three related findings.

First, need is not yet proven. Duke’s modeling shows significant load growth, but “much of the growth appears to be based upon anticipated data center customer additions,” the order states. “Any such anticipated load growth is insufficiently reliable for the Commission to act at this point.” Commission staff has not finished reviewing that forecast in the pending 2025 Carbon Plan and Integrated Resource Plan (CPIRP) case. Approving a half-billion-dollar unit first, the majority wrote, “will increase substantially the costs that ratepayers will need to bear and will impose a significant risk that new generation is approved before the need for that generation is determined.”

Second, the price is high for a peaker. Public Staff testimony called the $584 million tag “staggering” and “very expensive.” Commissioner Donald van der Vaart, concurring, said the record lacked a facility-specific comparison of alternatives, including battery storage that might meet near-term need at lower cost and with less stranded-asset risk. Commissioner Tommy Tucker cited the size of the expenditure and the possibility the unit could sit underused if data-center projects slip, shrink, or vanish.

Third, Duke did not show how this plant complies with the Ratepayer Protection Pledge it signed earlier in 2026. The White House pledge is voluntary. It asks utilities and large-load customers to keep households and small businesses from paying for generation and grid upgrades built primarily for data centers. The commission treated the pledge as relevant to the public-interest analysis even though it is not a statutory substitute for the CPCN standard. The record, the order said, “does not adequately establish the specific extent to which the Proposed Facility is intended to serve the anticipated growth in data center customer demand and therefore does not demonstrate how [Duke] intends to adhere to its commitments under the Ratepayer Protection Pledge with regard to the Proposed Facility.” If Duke refiles, it must offer cost-recovery mechanisms that meet that standard.

McKissick dissented. He argued the turbine is needed for reserve margin and reliability whether or not Amazon shows up, and that mixing the pledge into a CPCN case was “like mixing apples and oranges.” The majority was not persuaded.

Duke called the ruling disappointing and said it is assessing next steps. Spokesperson Craig Wilson said the company believes it has shown the unit is “part of a least-cost path to maintain reliable and affordable service for customers as energy demand continues to grow across North Carolina.” The denial does not kill the Amazon campus. Amazon and Duke already have air permits for hundreds of diesel generators that will bridge the site until it is fully interconnected. Those permits have drawn their own local opposition over particulate emissions.

Did Duke Make Enough Commitments to Shield Ratepayers?

Duke has made two overlapping sets of promises.

At the federal level, it signed the Ratepayer Protection Pledge. In North Carolina, it markets a “Customer Protection Plus” framework: engineering studies before interconnection, long-term electric service agreements, customer-funded connection costs, financial security, termination charges, and the claim that large loads contribute more revenue than they cost to serve. Duke has cited an analysis of 16 new large-load customers in the Carolinas that it says will cover their own costs and produce $3.6 billion in net value for other customers. On its Q2 2026 earnings call, the company said it has secured 7.8 GW of electric service agreements for data-center-type load.

That was not enough for this certificate. The commission wanted plant-specific answers: how much of this turbine’s output and capital cost is attributable to data centers; how those customers will pay (up-front contribution in aid of construction, special tariff, minimum-bill provisions, stranded-cost protections); and why this resource, at this price, beats alternatives on a constrained timeline. Governor Josh Stein and Attorney General Jeff Jackson had already told Duke that a Washington pledge is not a North Carolina tariff. They want a legally binding large-load tariff at the NCUC. That docket is still moving. Until it is finished, and until the CPIRP load forecast is tested, a $584 million peaker next to a hyperscale campus looks to the majority like a cost that could land on residential and small-business bills if the data-center megawatts do not materialize as modeled.

In short: Duke has frameworks and signed letters. It has not yet converted them, in this proceeding, into enforceable, unit-specific protections the commission is willing to accept.

Duke Energy’s Financial Position

The rejection lands against a strong corporate backdrop. Duke is a large, fully regulated utility holding company. It is not a distressed developer looking for a one-off offtaker.

Full-year 2025 results: reported and adjusted EPS of $6.31, up from reported $5.71 and adjusted $5.90 in 2024. Management introduced 2026 adjusted EPS guidance of $6.55 to $6.80 and extended its 5–7 percent long-term adjusted EPS growth rate through 2030, off the 2025 guidance midpoint of $6.30, with stated confidence in the top half of the range beginning in 2028 as contracted large-load growth ramps. The five-year capital plan stands at $103 billion, described by the company as the largest fully regulated program in the industry. Funds from operations to debt improved to about 14.8 percent, with a 2026 target near 14.5 percent and a longer-term goal of 15 percent.

Second-quarter 2026: reported EPS of $1.38 and adjusted EPS of $1.43, versus $1.25 in the year-earlier quarter. First-half 2026 reported EPS was $3.35 versus $3.00. Net income available to common stockholders was $1.077 billion in the quarter and $2.613 billion in the first half. Total operating revenues in the first half were $16.770 billion versus $15.757 billion. Electric Utilities and Infrastructure remains the earnings engine; growth is coming from rate-base investment and customer additions, offset by higher depreciation and interest on a larger asset base.

The company can absorb a delayed peaker. What it cannot easily absorb is a pattern of CPCN denials if data-center interconnection continues to outrun proven load and approved cost-allocation rules. Management has already flagged $5–10 billion of potential incremental capex beyond the current plan for generation and transmission in other jurisdictions. North Carolina is still the core franchise.

How North Carolina’s Power Prices Compare

North Carolina remains a relatively low-cost electricity state, which is part of why hyperscalers want to locate there. It is not, however, immune to the national updraft in rates.EIA Electric Power Monthly data for July 2026 put the average residential price in North Carolina at 15.16 cents per kWh, against a U.S. average of 18.31 cents. That is roughly 17 percent below the national average and ranks North Carolina near the mid-teens among the 50 states and D.C. (about 16th-lowest in that month’s snapshot). The June 2026 figure was 14.74 cents versus a U.S. average of 18.34 cents. Neighboring South Carolina was 15.47 cents in July; Virginia was 17.55 cents. The cheapest states clustered in the 13-cent range (Nevada, Nebraska, Utah, Louisiana). Hawaii, California, and much of New England remain two to three times higher.

The gap is narrowing on a year-over-year basis. North Carolina’s July residential rate rose 13.4 percent from 13.37 cents in July 2025. The national increase over the same window was about 4.9 percent. Higher-than-average usage in the Carolinas (roughly 1,300-plus kWh per residential customer in summer months versus a U.S. average near 1,160 kWh) means monthly bills can look closer to the national figure even when the unit rate is lower. An average July 2026 residential bill in North Carolina was reported near $203, versus a U.S. figure near $212. Five-year rate growth in the state is roughly in line with the national trend, in the low-30-percent range.

Duke has long argued that its rates sit below the national average and that large-load revenue can spread fixed costs. The commission’s order is a reminder that regulators will not take that argument on faith when the next increment of generation is a costly peaker justified largely by projects that have not yet signed final, fully protective contracts.

What Comes Next

Duke can refile after it shores up the demand case, the alternatives analysis, and the cost-allocation story. The CPIRP decision is expected later in 2026. A large-load tariff proceeding is already in motion. Amazon’s campus is still under construction. Temporary diesel generation will fill the gap in the near term.

The broader market context is the same one Oilprice and others have described: U.S. gas-fired capacity under development has surged, a large share of it tied to data centers, and state commissions are beginning to ask who pays if the electrons do not show up. North Carolina just became one of the first GOP-led commissions to say the answer has to be on the page before the shovel goes in the ground.

The Smith turbine is not dead. The assumption that a signed federal pledge plus a corporate “protection plus” slide deck equals a CPCN is.

Making Appendices Great Again

Check out the World’s Greatest Podcast Show Notes at EnergyNewsBeat.co or EnergyNewsBeat.com.