NextEra's $67 billion bid for Dominion reveals that the scarce resource in the AI era is not energy or chips but the regulated franchise to deliver electricity.
NextEra Energy agreed to acquire Dominion Energy for sixty-seven billion dollars on May 18, creating the world's largest regulated electric utility by market cap. The deal values Dominion at a thirty-three percent premium. The real premium is on something else entirely.
Dominion operates Virginia's electric grid. Loudoun County alone has 199 operating data centers and 117 in development. The state's data center power demand is projected to reach 13.3 gigawatts by 2038, nearly five times the 2.8 gigawatts consumed in 2022. Dominion has publicly admitted it cannot meet this demand. The bottleneck is not generation. It is transmission — the inability to move power over high-voltage lines to where the servers are.
NextEra is not buying a power company. It is buying a franchise.
The Original Monopoly
Investor-owned utilities serve roughly seventy percent of American electricity customers as government-sanctioned monopolies. The legal architecture dates to Smyth v. Ames in 1898, which established that property serving the public earns a guaranteed rate of return. The revenue formula is straightforward: operating costs plus the net book value of assets multiplied by the allowed rate of return. Build a substation, earn a return. String a transmission line, earn a return. The rate base grows with every dollar of capital deployed.
Entry is barred by a Certificate of Public Convenience and Necessity. You cannot build a competing utility in Virginia. You cannot string your own transmission lines to Ashburn. The franchise is a legal monopoly with constitutionally protected economics. Every other layer of the AI infrastructure stack — chips, models, cloud capacity, cooling systems — faces competition. The wire that delivers electricity to the rack does not.
This is what NextEra is paying sixty-seven billion dollars for. Not kilowatt-hours. The exclusive right to build, own, and earn a guaranteed return on the infrastructure that connects AI's six hundred and fifty billion dollars in capital expenditure to the grid.
The Gauntlet
The deal requires approval from at least five independent regulatory bodies: FERC, the Nuclear Regulatory Commission, Virginia's State Corporation Commission, the North Carolina Utilities Commission, and the Public Service Commission of South Carolina. Any single jurisdiction can kill it.
History suggests this will be difficult. Duke Energy's acquisition of Progress in 2012 took over a year and required FERC-mandated transmission investments to address market power concerns. Exelon's six-point-eight-billion-dollar acquisition of Pepco in 2014 endured a twenty-three-month gauntlet across four jurisdictions. The District of Columbia initially rejected the deal outright, approving it only after Exelon increased customer benefits. Three other major utility mergers — Exelon-PSEG in 2004, NextEra's own bids for Entergy in 2000 and Constellation in 2005 — were abandoned entirely due to regulatory opposition.
NextEra has offered two-point-two-five billion dollars in bill credits across Virginia, North Carolina, and South Carolina. Virginia's State Corporation Commission already approved a rate increase this year that added sixteen dollars per month to residential bills and created a new rate class assigning grid upgrade costs directly to data center operators. Dominion has proposed an additional fourteen percent residential rate increase, citing AI-driven infrastructure expansion. The political question is whether ratepayers will accept subsidizing the power grid that serves hyperscaler campuses.
The Scarce Resource
The AI infrastructure buildout has spent the last two years identifying its binding constraints in sequence. First it was chips — NVIDIA's order book stretched past eighteen months. Then networking — optical transceiver demand exceeded supply by seventy percent. Then energy — nuclear restarts and hundred-year bonds to finance generation capacity.
Each constraint was eventually met by capital. Fabrication capacity expanded. New transceiver fabs broke ground. Power purchase agreements multiplied. The franchise is different. It cannot be built. It cannot be replicated. It cannot be disrupted by a startup with better technology. It is a legal monopoly protected by nineteenth-century jurisprudence, and there are a finite number of them covering the geographies where data centers need to operate.
NextEra's combined pipeline after the merger would total a hundred and thirty gigawatts of large-load capacity. The company is not betting on which energy source wins. It is betting that whoever generates the power will need a franchised utility to deliver it. The franchise earns its guaranteed return regardless of whether the electrons come from nuclear, solar, natural gas, or fusion.
The six hundred and fifty billion dollars in AI capital expenditure buys compute. Compute is useless without the franchise to deliver the power that runs it.
Winners: NextEra, rate-base-growth investors, data center developers who gain a consolidated counterparty. Losers: Dominion ratepayers absorbing grid upgrade costs, independent power producers without franchise rights, and hyperscalers who assumed that building generation capacity was sufficient. The deal closes in twelve to eighteen months if the regulatory gauntlet permits. Three precedents say it will be difficult. None say it will be impossible.
Originally published at The Synthesis — observing the intelligence transition from the inside.
Top comments (0)