But the more valuable insight sits one step past the dockets: this regulatory arc has a known shape, it has played out in other industries before, and knowing where it ends tells you exactly when to move. The answer is not “eventually.” The answer is now, and for a reason most teams have not articulated.
The record, in sequence
The turn has a specific timeline, and it reads like escalation because it is. In December 2025, FERC issued an order finding PJM's tariff unjust and unreasonable for its lack of clarity on co-location arrangements, directing the nation's largest grid operator to establish new transmission services for co-located load, clarify interconnection procedures, and revise its behind-the-meter generation rules. Regulators do not deploy the phrase “unjust and unreasonable” against rules they intend to preserve. That order was the turn.
PJM's stakeholder process produced the operating architecture in January 2026: an expedited Bring Your Own Generation track, a 50 MW threshold defining large loads, improved load forecasting. The compliance filing followed in February 2026, proposing revised behind-the-meter rules and three new transmission services for co-located loads, with a requested effective date of July 31, 2026. A buyer designing a co-located facility today is designing into a named, sanctioned, dated track, not improvising against regulatory silence.
The June 2026 orders nationalized the logic. FERC directed six grid operators, PJM, MISO, SPP, CAISO, ISO-NE, and NYISO, to either justify their existing tariffs or file reforms across five areas, explicitly including accommodation of co-location and behind-the-meter generation and new services for flexible large loads, with the commission stating it will not hesitate to dictate solutions if the filings come back light. Across FERC, DOE, and the White House, the federal posture moved from rejection to active rulemaking inside a single procurement cycle, and the commission now brands the effort in the language of speed to power.
The second act: cost allocation, not permission
The permission question is functionally settled; the fight beginning now is over who pays for the grid, and large loads are being fitted for the villain costume. The price signal is already on the record, and it is violent. PJM's capacity auction cleared at $28.92 per MW-day for the 2024/25 delivery year. The 2025/26 auction cleared at $269.92, a ninefold jump, with the BGE and Dominion zones clearing at $466.35 and $444.26. The next two auctions pinned the FERC-approved price cap, at $329.17 and then $333.44 per MW-day, with the 2027/28 auction also falling 6,623 MW short of PJM's reliability requirement, and annual capacity spend across the footprint now runs on the order of $16 billion. Three consecutive auctions at the ceiling, a near-tenfold repricing from the 2024/25 baseline, and a formally declared reliability shortfall: that is the arithmetic ratepayer advocates are carrying into every rate case, and data center load growth is the named cause in most of the filings.
Electricity affordability has become the dominant utility-sector concern, and Washington is responding with instruments aimed squarely at how data center costs are allocated. Every rate case in every growth market for the next five years will feature a version of the same question: did the data center pay its own way?
Read behind-the-meter generation through that lens and its strategic value doubles. The load that brings its own generation is the load no rate case can blame, and the load exposed to none of those capacity prices. On-site power is not just the fastest path around the queue documented in The Queue Is the Product; it is political insulation in an environment where the alternative is being the named defendant in the affordability debate. Utility-sector research has already observed data centers pursuing on-site power precisely as affordability tops the industry's concerns. Teams treating BTM purely as a schedule instrument are capturing half its value.
The third act is visible too: the workaround gets regulated
Now extend the arc one act further than the dockets go, because regulatory history runs on a loop and this loop is well documented. Every workaround that succeeds becomes a category; every category that grows becomes a constituency; every constituency large enough to matter becomes a regulated class. BTM is being mainstreamed, which means BTM will eventually be metered, tariffed, and cost-allocated in its own right: standby charges, grid-services obligations, exit fees, contribution requirements. Not because regulators are hostile, but because a workaround that carries twenty gigawatts is no longer a workaround, it is infrastructure, and infrastructure pays. The probability-weighted sequence: permission settled now, rate design fought through 2027, BTM-specific obligations arriving as BTM load share becomes material, on the other side of that.
And for anyone who thinks the obligation phase is speculative, it has already started on the grid-served side of the line. AEP Ohio's approved data center tariff requires large loads to commit to paying for the substantial majority of their subscribed capacity, on the order of 85 percent, under contract terms running a decade and more, with exit fees and grandfathering written into the same order, and other utilities have filed in the same direction. Grid-served large loads are already being fitted with long-tenor minimum-take obligations. That is what the settled regime looks like, arriving on schedule.
This is why grandfathering is not a procedural footnote; it is the entire game. FERC's PJM order already requires a transition period and the grandfathering of certain existing contracts, which is how these regimes always settle. Think of regulatory position the way fund investors think of vintages: arrangements executed in the 2026-2027 window, while the rules are forming, are the cheapest regulatory vintage that will ever exist in this category. They get protected status, they anchor the baselines against which later obligations are measured, and they are negotiated against regulators who still need early movers to validate the framework. Arrangements executed after settlement inherit whatever cost allocations and contested compromises the paper hearings produce, negotiated against regulators who no longer need anyone. Early movers get the regime they negotiated. Late movers get the regime everyone else negotiated for them, and the AEP tariff is the preview. The same vintage logic prices the grandfathered turbine positions described in The Turbine Gap, and it is not a coincidence.
In every part of this market, the paper written before the rules settle is the paper worth holding.
The honest caution
The rules are directionally settled and specifically unfinished. Rates, terms, and conditions for the new transmission services are being determined through paper hearings, several operators can take extensions, and the complete national picture likely lands between late 2026 and early 2027. Grandfathering scope and cost-allocation details will be fought line by line. Directional certainty is not a signed service agreement, and nothing here substitutes for counsel in your specific market.
Operationally, the risk conversation has inverted twice. Two years ago the question was whether a co-located facility would be allowed to operate. Today the question is which sanctioned configuration fits your load and how fast you qualify for the expedited track. And the question after that, the one this article exists to put on your desk early, is whether your arrangements will hold protected vintage when the category matures. Organizations still litigating the permission question internally are spending the one asset this market runs on, which is time, to re-answer a question the record closed, while the window that actually matters, the vintage window, quietly runs.