FERC didn't pick a side in the data-center fight. It picked a better question.

The most satisfying thing you can say about data centers and your electricity bill is that the data centers are the problem and someone should make them stop. It is satisfying because it is partly true: large new loads are arriving on a grid that was not planned for them, and somebody is going to pay for the wires and the generation that connect them. When a city council in Seattle votes nine to nothing for a moratorium, or a state legislature starts drafting a ban, it is responding to a real grievance with the bluntest instrument in the drawer.

The opposite position is just as popular in different rooms. There, the problem is the queue — the multi-year wait to interconnect, the studies that pile up, the rules written for a slower era now strangling the most important industrial buildout in a generation. Get out of the way, this version says. The demand is real, the reliability case is real, and every month of delay is a month of capacity that gets built somewhere with a friendlier regulator.

On June 18 the Federal Energy Regulatory Commission did neither of these things, and I think it was right not to. It declined to answer the question everyone is shouting — should the data centers connect — and asked a quieter, more durable one: on what terms, and who pays. Then it routed that question through one of the least glamorous and most powerful tools in administrative law. I say this as someone who used to draft the kind of order I am about to defend, which means I also know exactly how it can still go wrong.

Both of the loud positions deserve a fair hearing

Let me steelman the backlash first, because it is not irrational and the people making it are not fools. Electricity is a shared system with shared costs. When a hyperscaler shows up asking for a gigawatt, the grid has to be expanded to serve it, and under a lot of existing tariffs the cost of that expansion gets socialized — spread across every ratepayer's bill — even though the new load is the reason it was incurred. If your power bill is rising and the new substation down the road is feeding a building with no windows and forty security cameras, you do not need a degree in rate design to feel that something is off. You are, in a narrow and real sense, subsidizing a trillion-dollar industry's input costs. That is a legitimate complaint, and the moratorium is what a legitimate complaint sounds like when it has no better channel.

Now the other side, also stated fairly. The interconnection process in much of the country genuinely is too slow, and slowness is not free — it is reliability risk deferred and investment pushed offshore. The grid operators are being asked to absorb demand growth that no forecasting model saw coming, and the honest ones will tell you the existing rules were built for a world where load grew at one or two percent a year, not for one where a single customer can ask for the output of a mid-sized power plant. Telling that system to simply go faster, without telling it how to allocate the cost, is its own kind of magical thinking.

Both positions are responding to the same fact — the load is here and the rules don't fit it — and both reach for the same shape of answer: a yes or a no. Let them in. Keep them out. That shared instinct is the category error, and it is worth naming precisely, because nearly every bad rule in this fight will descend from it.

What the commission actually did

FERC issued what are called show-cause orders, under Section 206 of the Federal Power Act, to all six of the regional grid operators it oversees: PJM, MISO, SPP, CAISO, ISO New England, and the New York ISO. Together they run the wholesale grid for something like 200 million Americans across more than 30 states. The vote was unanimous. The legal move underneath it is the part worth slowing down for, because the procedure is the policy.

Most of the time, when a regulator wants to change the rules, it proposes the new rule and invites comment — it carries the burden of justifying the change. Section 206 lets the commission do something closer to the opposite. By finding that the existing tariffs 'appear to be unjust and unreasonable' for handling large-load interconnection, FERC shifted the burden onto the grid operators: defend your current rules, on the record, or reform them. The operators have 60 days to respond with briefing and answers to specific questions, and 30 days to file reports on whether they can actually secure enough generation to keep the lights on as these loads arrive. The commission also named five areas it expects the answers to address — transmission-service and alternative-technology options; cost-shifting and cost transparency; co-location and behind-the-meter generation; flexible load that can power down when the grid is tight; and how to study power plants built to serve a single large customer.

Read that list again and notice what is not on it. There is no number for how many data centers may connect. There is no approval gate, no national cap, no ban. Every item is a question about terms — about who bears which cost, under what conditions, with how much disclosure. FERC looked at a fight that everyone wants to be about permission and insisted it is about design.

FERC looked at a fight everyone wants to be about permission, and insisted it is about design. — Vic Reyes

The better question, and why it is unsatisfying

Here is the frame I think is correct, and I will admit up front that it pleases no one. The grid does not have a data-center problem. It has a cost-allocation problem that data centers happen to have made impossible to ignore. The right question is not whether a given load gets to exist; it is whether the entity that causes a cost is the entity that pays it. Make the load that triggers a hundred-million-dollar upgrade carry that upgrade's cost, and most of the public's grievance dissolves, because the thing people actually resent is not the building — it is the subsidy hidden in their bill. Fail to do that, and no moratorium will fix it, because the next data center will simply be built one jurisdiction over and the cost will find its way back to a shared ledger somewhere.

This is duller than a ban and it is also the only version that survives contact with how the grid is actually run. A moratorium stops one building; it does nothing about the rule that mispriced the building in the first place. A burden-shifting order aimed at the rule is slower, harder to explain, and impossible to put on a placard. It is also the move most likely to change the thing that is actually broken. Good regulation is frequently the boring choice that addresses the real mechanism instead of the satisfying choice that addresses the symptom.

Who actually enforces this

Now the question I learned to ask before believing any order, including ones I wrote: who enforces this, and how. And here the FERC action runs straight into the seam that makes American energy regulation so genuinely hard. FERC's jurisdiction is wholesale — interstate transmission and wholesale power markets. It can guard against cost-shifting among transmission customers. But the bill that lands in your mailbox is a retail rate, and retail rates are set by your state commission. FERC can tell the grid operators to stop socializing transmission costs onto the wrong parties; it cannot, by itself, write the rule that decides what a residential customer in Ohio or Georgia ultimately pays.

That is not a flaw in the order so much as the terrain it has to cross, and the smarter states are already moving on their half. Virginia, where the data-center density is highest in the country, has been building a rate structure that makes large customers carry the lion's share of the demand-related cost they cause — on the order of 85 percent of transmission and distribution demand and 60 percent of generation demand, rather than letting it bleed onto everyone else. That is the retail-side complement to what FERC just did on the wholesale side, and the two only add up to ratepayer protection if both halves actually get built. One without the other is a half-finished bridge.

This is also why FERC's decision to issue six tailored regional orders, rather than one national rule, is shrewder than it looks. The National Association of Regulatory Utility Commissioners — the state regulators — had warned that heavy federal standardization would trample state authority and slow the local adaptation this problem needs. By writing region-specific orders and explicitly preserving state jurisdiction over retail and distribution, FERC did two things at once: it respected a real federalism boundary, and it gave its own action a far better chance of surviving the legal challenge that uniform national mandates reliably attract. A regulator who has been sued knows that the most elegant rule is worthless if a court vacates it.

The line that gives the game away

I want to flag the sentence in the chair's statement that I keep returning to, because it is where the order's ambition meets its tension. Alongside the language about a resilient grid and protected consumers, Chairman Laura Swett emphasized providing 'certainty for investors' and directing the markets to 'protect existing deals.' I understand why that is there. Capital will not build transmission and generation into a regulatory environment that might retroactively rewrite the economics, and you cannot solve a capacity problem while spooking the people who fund capacity.

But hold 'protect existing deals' next to 'prevent cost-shifting' and you can see where the friction lives. Some of the existing deals are the cost-shifting. If a data-center interconnection was signed under a tariff that socialized its upgrade costs, then protecting that deal and ending that subsidy are, in at least some cases, the same sentence asking for two different things. The order does not resolve this; it cannot, at the show-cause stage. But it is the place to watch. A reform that grandfathers every arrangement already on the books would calm investors and quietly preserve the exact transfer that has people showing up at city-council meetings. Who is made whole, and who is grandfathered, is where this stops being a press release and becomes a policy.

The thing I would worry about is the clock

If I have a real fear about this action, it is not the direction. It is the speed. FERC compressed what would ordinarily be a multi-year rulemaking into 30- and 60-day windows. I am sympathetic to the urgency — the demand is not waiting, and a regulator that moves at its usual pace would be writing rules for a grid that has already been reshaped without it. But I have seen what happens when good intentions meet an impossible deadline inside an agency. You get filings assembled in a hurry, reform proposals that are really just defenses of the status quo with new formatting, and a comment period too short for the parties without a full-time Washington team — which, not coincidentally, means the ratepayer advocates and the small intervenors, the very people the order is supposed to protect. Rushed process tends to favor whoever already has the lawyers in the building.

So the burden-shift is the right instrument, and the calendar is the part that could waste it. The honest version of my position is that Section 206 was the correct tool and that its value will be decided almost entirely by whether the commission treats these deadlines as a forcing function or a finish line. If the 60 days produce a genuine record and a real second round, this works. If they produce a box-checking exercise that ratifies what the grid operators were going to do anyway, FERC will have spent its rarest authority to move a problem from one in-tray to another.

The frame, with its costs admitted

Here is what I would tell the version of me still inside the building. The instinct to answer the data-center question with a yes or a no is the thing to resist, in both its populist and its libertarian dress. The grid is not a velvet rope. It is a cost-allocation machine, and the only durable fix is to make the machine charge the cost to the load that caused it — at the wholesale seam where FERC sits, and at the retail seam where the states do, with both halves actually completed and neither one used as an excuse to skip the other.

That frame has costs, and I am not going to pretend otherwise. It is slow when people want fast. It is technical when people want a villain. It will leave a patchwork at the edges, because a federated grid governed by one federal commission and dozens of state ones cannot produce a single clean rule, and anyone promising you it can is selling the category error in a nicer suit. What it offers in return is the only thing that actually lasts: a system where the people angry about their bills have a real reason not to be, and the people building the future are not doing it on a subsidy nobody voted for. FERC did not end this fight on June 18. It did something rarer. It pointed the fight at the right question, and then started the clock on whether anyone can answer it in time.