Americans want the benefits of AI, but not the machinery. Gallup now finds 71% oppose an AI data center in their area, making local server farms more politically toxic than nuclear plants. That is a remarkable achievement for a building whose main offense is using electricity, water, and enough concrete to make a county commissioner feel briefly important.

The deeper story is not just NIMBYism. PJM is admitting the grid has moved from surplus to scarcity, FERC’s chairman is questioning whether the nation’s largest power market is governable, and Eversource is openly resisting data centers to protect customers from higher prices. The AI boom has finally reached the physical world, where the future must compete with substations, gas turbines, transmission studies, utility bills, and the ancient American right to hate whatever is being built nearby.

NextEra Energy plans to acquire Dominion Energy in an all-stock deal that would create what the companies describe as the world’s largest regulated electric utility by market capitalization. The combined company would serve about 10 million utility customers across Florida, Virginia, North Carolina, and South Carolina, with a $138 billion rate base and more than 130 GW of large-load opportunities in its pipeline. Dominion shareholders would receive 0.8138 shares of NextEra for each Dominion share, and the companies expect the deal to close in 12 to 18 months if regulators approve it.

The deal is a data center story wearing a merger suit. Dominion sits in Virginia’s data center alley and recently said its contracted data center capacity pipeline stands around 51 GW. NextEra brings one of the country’s largest generation development platforms, including gas, renewables, nuclear, and storage. Analysts immediately framed the merger as a possible return toward the integrated utility model, especially in PJM, where resource adequacy is becoming less a market-design problem and more a civilization-maintenance hobby.

Grid Take: The market is starting to reward utilities that can offer large customers a complete answer: power, wires, financing, permitting, and speed. If the AI load boom is real, the winning utility model may look less like a passive wires company and more like an infrastructure machine with a balance sheet.

The Pennsylvania Public Utility Commission released a “first-of-its-kind” model tariff framework for large-load customers, including data centers. The order is nonbinding, but it recommends applying large-load provisions to customers exceeding 50 MW individually or 100 MW in aggregate, and says utilities should charge those customers for system upgrades that would not have been needed “but for” their interconnection, even if other customers eventually benefit from the infrastructure.

That is the sentence every ratepayer advocate has been waiting to hear. The framework also covers collateral, deposits, stranded-asset risk, load ramping, minimum contract terms, exit provisions, and public-facing information on interconnection requests. It even allows large-load customers to self-construct some upgrades, which is a useful concession to the obvious reality that hyperscalers can often move faster than the utility process, a machine built partly from paper, partly from caution, and partly from the fossilized remains of 1970s hearing rooms.

Grid Take: Pennsylvania is trying to separate “economic development” from “please make grandma finance my substation.” That is exactly where the data center fight has to go next: not yes or no, but yes under terms that protect existing customers and still let serious projects build.

EIA now expects commercial electricity consumption to surpass residential use for the first time on record in 2027. The commercial sector, which includes hyperscalers, bitcoin miners, and cloud computing, is expected to reach about 1,530 billion kWh in 2026, roughly equal to residential consumption, before growing another 5.3% in 2027. Total U.S. electricity consumption is expected to rise 1.3% in 2026 and 3.1% in 2027.

Residential demand, by contrast, is expected to remain mostly flat, growing about 0.5% in both 2026 and 2027. But residential customers will still feel the shift. EIA expects average residential prices to rise to 18.2 cents/kWh this year, nearly 5% above 2025, with another 2% increase expected next year. The East Coast is expected to see some of the largest increases, with utilities citing fuel costs, transmission investment, extreme-weather hardening, and rising demand.

Grid Take: This is the quiet statistical handoff that explains the whole decade. The American grid was built around households and legacy industry; it is now being reorganized around compute, manufacturing, electrification, and commercial load that arrives in chunks large enough to make planners blink twice.

EPA proposed revisions to wastewater rules for coal-fired power plants, including rescinding treatment requirements tied to unmanaged combustion residual leachate from coal ash disposal sites. The agency argues the rollback could reduce electricity-generation costs by up to $1.1 billion annually and says growing electricity demand from AI infrastructure and data centers “cannot be met under the overly restrictive policies of past administrations.”

Critics, including Earthjustice, say the change would eliminate safeguards on wastewater containing arsenic, mercury, and other dangerous contaminants. That makes this a revealing story less because of the specific wastewater rule and more because of the political logic now forming around AI load growth. Demand is becoming the universal solvent in energy policy, used to justify more generation, slower retirements, faster permitting, regulatory rollbacks, and probably, by next Tuesday, a county commissioner’s nephew getting a consulting contract.

Grid Take: The AI power crunch is going to reopen fights people thought were settled, including coal regulations, gas permitting, transmission siting, nuclear licensing, and water use. The danger is that “we need more power” becomes either a blank check for bad policy or a veto against building anything at all.

The National Taxpayers Union released Speed to Power: How Electricity Ratepayers Can Win the AI Race, a wide-ranging policy guide from Nick Loris with contributions from Travis Fisher, Josh Smith, Katherine Wright, and Pete Sepp. The report argues that large data center demand can strain the grid, but it can also reduce residential rates by spreading fixed utility costs across a broader customer base and underwriting upgrades that benefit the shared system, if policy gets the incentives right.

The strongest parts of the paper are its practical framework: permitting reform, fair cost allocation, better load forecasting, more competition, data center flexibility, water-rights reform, and a proposal for Consumer Regulated Electricity, which would allow privately financed, physically islanded power systems to serve voluntary customers like data centers or industrial facilities. It is a useful antidote to the two lazy positions now dominating the debate: pretend data centers are free money, or treat them as invading machinery from a bad science-fiction sequel.

Grid Take:
The NTU paper gets the core question right: data centers are not automatically a subsidy scam or a ratepayer miracle. They become one or the other depending on whether policymakers force accurate cost causation, protect existing customers, and let private capital build where the legacy grid cannot move fast enough.

Upgrade to Grid Brief Premium to get extra deep dives into energy issues all over the world.

We rely on word of mouth to grow. If you're enjoying this, don't forget to forward Grid Brief to your friends and ask them to subscribe!