Don't Approve the NextEra-Dominion Mega-Merger on AI's Word Alone

On Monday morning, NextEra Energy and Dominion Energy announced a $67 billion all-stock combination that would create the largest regulated electric utility company in the world. Roughly ten million households and businesses across Florida, Virginia, North Carolina, and South Carolina would be served by the merged firm. NextEra chairman and chief executive John Ketchum, who would lead the combined company, told analysts the deal is necessary to build the power projects “hyperscalers” — Amazon, Microsoft, Google, Meta — are demanding fast enough and cheaply enough to meet them. Dominion chief executive Bob Blue would run the combined regulated-utility business. The proposed close is mid-to-late 2027, contingent on approval from the Federal Energy Regulatory Commission, the Nuclear Regulatory Commission, and the state utility commissions of Virginia, North Carolina, and South Carolina.

The companies are offering Dominion ratepayers in the three Atlantic states a total of $2.25 billion in bill credits, spread over two years after the deal closes. That works out to roughly $15 a month for an average residential customer for twenty-four months. Then the credit stops, and a much larger company that no state regulator has ever supervised before runs the grid for the next several decades.

This newspaper’s position is straightforward: the merger should not be approved on the strength of a data-center demand forecast and a two-year bill credit. The deal asks regulators in four states to bind ten million customers to a single corporate structure on the assumption that AI electricity demand will grow as projected, that a larger utility will serve them more cheaply than a smaller one, and that the combined firm will resist the inevitable temptation to load capital costs onto residential ratepayers and small businesses. Each of those three assumptions deserves rigorous, public, line-item scrutiny before a single regulator signs.

The AI demand story is the whole pitch

Read the joint announcement and the SEC 8-K filing carefully and almost every claimed customer benefit traces back to the same load forecast. The combined company will be able to build generation faster, the executives say, because it will have a unified planning function across four states. It will procure equipment more cheaply because of scale. It will negotiate better terms with hyperscaler customers because it owns more of the relevant grid. The strategic logic is, in Ketchum’s framing, that data-center demand is so large and so urgent that only a regulated utility of unprecedented size can build to meet it.

The forecast itself is real. PJM Interconnection, the regional grid operator that includes Dominion’s Virginia service territory, projected in its 2026 Long-Term Load Forecast that its Dominion zone will see the largest absolute increase in summer peak demand of any zone between 2026 and 2030. Summer peak load in that zone was 23,905 megawatts in 2025, already 23 percent above 2019. The U.S. Energy Information Administration has documented that commercial electricity sales in Virginia have grown faster than in any other state in the country, driven almost entirely by data centers in Loudoun and Prince William counties. The Institute for Energy Economics and Financial Analysis estimates PJM-wide capacity-market costs will rise roughly $1.4 billion annually starting in June 2026, with the increase tied “largely to data center demand.”

The numbers are not in dispute. What is in dispute is who pays for the buildout those numbers imply, and whether the proposed corporate structure is the right vehicle for paying it. On both questions, the merger announcement is silent.

Two years of credits, decades of capital recovery

State utility regulators in Virginia, North Carolina, and South Carolina set the rates Dominion charges. Their job, under the regulatory compact that has governed American electric utilities for a century, is to set rates high enough that a prudently managed utility earns a reasonable return on capital invested in serving customers — and no higher. Every dollar of authorized capital is, over time, a dollar repaid by ratepayers, with interest, at a regulator-set rate of return.

Capital investment of the scale Ketchum is describing — building new gas plants, new transmission, possibly new nuclear, to serve hyperscalers in Northern Virginia and elsewhere — does not get paid for out of shareholder cash. It gets put into a rate base, and ratepayers pay it back over a generation. The relevant question for regulators is not whether the combined company will build the infrastructure. It is whether the structural design of the combined company makes it likelier or less likely that the cost of that infrastructure ends up on the wrong customer’s bill.

The $2.25 billion in announced bill credits is real money. It is also less than 4 percent of the announced deal value, paid out over two years, in a service area where residential rates have been climbing on the back of capacity-market spikes and where the Virginia State Corporation Commission only last November approved a new rate class specifically to wall off data-center demand from residential bills. The credit gets a customer through roughly the first year and a half of the merger. The merger lasts decades.

Virginia’s new large-load rate class, approved in 2025, requires affected customers (predominantly AI data centers) to pay for at least 85 percent of contracted distribution and transmission demand and 60 percent of generation demand starting in January 2027. That rule exists because Virginia regulators, on their own initiative, concluded that data-center load was big enough and growing fast enough that the historical regulatory presumption — that all customers benefit roughly proportionally from system upgrades — was no longer safe. The merger announcement does not address this rule. It does not commit the combined company to defending it. It does not commit the combined company to applying the same protection in North Carolina, South Carolina, or Florida. It buys two years of bill credits and asks regulators to take the rest on faith.

A bigger utility is not automatically a cheaper one

Ketchum’s pitch — that scale, by itself, makes power cheaper — is the oldest and most reflexive argument in the history of American regulated utilities, and it has a mixed record. Coverage of the deal by the watchdog group Common Dreams cited Public Citizen energy program director Tyson Slocum calling the proposal “absurd” and saying it “should be dead on arrival for state and federal regulators.” Clean Virginia executive director Brennan Gilmore, also quoted in coverage, pointed to what he called NextEra’s “deeply troubling track record” of rate disputes and political-spending controversies in Florida. Both groups are advocacy organizations with stated positions, and their language is sharper than the language of regulators. But their underlying point is the one that regulators are obligated to test directly: does the proposed combination actually produce the cost reductions Ketchum is promising, and does it deliver them to small ratepayers rather than to shareholders and hyperscaler customers?

That question can be answered on the record. The Federal Energy Regulatory Commission, FERC, has a long history of merger-review proceedings in which the applicants must prove, with line-item evidence, that synergies are real, that competitive harm is limited, and that rate effects on captive customers are protected. The three affected state commissions — Virginia, North Carolina, and South Carolina — have their own merger-review traditions, and each has previously rejected or extracted concessions from large utility combinations that did not meet the burden. The Nuclear Regulatory Commission’s role is narrower but real, because the merger involves nuclear ownership across multiple jurisdictions.

The right outcome here is not a reflexive no. It is a slow, public, evidence-based review that puts the burden of proof on the applicants and that treats the AI demand forecast as a starting point for questions rather than as a self-evident justification.

The strongest argument for the deal

The honest case for letting the merger proceed, on terms, is one the companies make less directly than they should. Building generation infrastructure at the speed the load forecast implies is genuinely difficult. The Virginia grid does need to accommodate substantial new load over the next decade. A larger, better-capitalized utility — particularly one with deep nuclear and renewables expertise from NextEra’s Florida operations — may, on the right terms, be more capable of executing that buildout than two smaller firms operating across a state boundary. The combined balance sheet is bigger. The planning function is unified. The procurement leverage is real. None of this is empty talk; it is the basic logic of why regulated monopolies exist in the first place.

Set against that, however, is the cost of getting this wrong. A merger of this size is, in practice, irreversible. Once approved, the combined company will run the grid for ten million households for the foreseeable future. The harms regulators failed to anticipate in approving previous mega-mergers — declining service reliability in some, opaque cost allocation in others, political capture in still others — are documented in the trade and academic literature. The argument for caution is the asymmetry: a slow review that costs the companies a year of integration timeline is recoverable. A bad approval that locks in twenty years of mis-aligned incentives is not.

What approval should look like, if it comes

If FERC and the three state commissions ultimately approve this deal, the approval should not look like the announcement. It should include, at minimum, four things the applicants have not yet offered. First, an enforceable, long-duration commitment to extending Virginia’s large-load rate-class protections — the residential firewall — into North Carolina, South Carolina, and any future Florida data-center growth, so that hyperscalers pay for the infrastructure built to serve them. Second, ratepayer-credit terms substantially larger than the announced $2.25 billion, scaled to the actual investment to be recovered from ratepayers, and structured so that the credit applies over the full period during which merger-driven capital is being recovered, not just the first twenty-four months. Third, public, audited reporting requirements on the synergies the merger is promised to deliver, with claw-back authority if those synergies are not materialized into rate reductions on a defined timetable. Fourth, divestiture and behavioral commitments adequate to address market-concentration concerns at FERC, including in capacity markets where Dominion-NextEra would become the dominant counterparty for a generation of new dispatchable supply.

None of those four conditions are radical. Each has precedent in prior FERC and state-commission merger reviews. Each can be quantified, monitored, and enforced. The absence of any of them from Monday’s announcement is the most informative thing about the announcement.

Across our recent coverage, the same pattern keeps repeating: huge, capital-intensive infrastructure designed to serve AI computation, approved by elected and appointed officials who acknowledge they could not really say no, in jurisdictions whose residents will pay for the buildout long after the news cycle has moved on. Taiwan Semiconductor’s most recent earnings report drove home the same constraint from the supply side: AI capacity is bottlenecked by electricity, not chips. The shape of the bottleneck — who builds the power, who pays for it, and who decides — is the most consequential domestic-economic question of this decade, and it is being settled, deal by deal, in regulatory proceedings that get a tenth of the coverage of the merger announcements themselves.

That is the case for slowing this one down. The companies have made their offer. The right answer from FERC and from regulators in Richmond, Raleigh, and Columbia is not to say yes, and it is not to say no on principle. It is to ask, on the record, the questions Ketchum’s announcement did not. Who pays for the data-center buildout, exactly, and in what proportion? What enforceable commitments will protect captive residential customers in all four states, not just for two years but for the life of the combined company’s capital recovery? What is the evidence — not the projection — that a larger utility will run a cheaper grid for the small customer? Until those answers exist in writing, on a docket, signed by people who can be held to them, the burden of proof has not been met. And until the burden has been met, the merger should not be approved.