The fast lane to nowhere? What FERC's June orders actually fix

On June 18, 2026, the Federal Energy Regulatory Commission did something it almost never does: it moved fast. FERC directed each of the six regional grid operators it regulates, along with their member transmission owners, to justify or reform the terms of their tariffs related to interconnections for data centers and other large energy users. The show cause orders, issued under Section 206 of the Federal Power Act, arrived with deadlines measured in weeks, not years. For an agency whose signature reforms typically gestate through multi-year rulemakings, this was a statement.

The statement was necessary. The United States is trying to build the physical layer of the AI economy on a grid whose front door has been jammed for a decade. The constraint on new AI capacity is no longer chips, capital, or land; it is the queue of projects waiting for permission to plug in. FERC's orders are the most consequential attempt yet to clear that queue.

But there is a distinction that much of the coverage has missed. The orders accelerate process: study timelines, tariff clarity, filing deadlines. They do not, and cannot, conjure generation. A faster queue that leads to the same insufficient supply of electrons is a fast lane to nowhere. The reform succeeds only if the generation pipeline moves through the widened gate. That is the test, and the early evidence suggests it will be a close call.

The backlog is structural, not administrative

Start with the scale of the problem FERC is trying to solve. The interconnection queue is not a waiting room; it has become larger than the building it leads into. As of the end of 2025, roughly 8,200 projects were actively seeking grid interconnection in the United States, representing 1,312 GW of generation and approximately 749 GW of storage. Combined, that is more than 2,000 GW of capacity in line, a figure that exceeds the entire installed generating fleet of the country.

A queue of that size is not a symptom of slow paperwork; it is a symptom of a system designed for a different era. The interconnection process was built for a world in which a handful of large, predictable plants requested connection each year. It now faces thousands of applicants, and it responds the way any overloaded system does: with attrition. Historical completion rates are brutal. Of all capacity that submitted interconnection requests between 2000 and 2019, only 13% had reached commercial operation by the end of 2024. The overwhelming majority of projects that enter the queue never leave it as operating assets.

This is the context in which FERC acted, and it explains why the Commission concluded that incremental tinkering had run its course. When a process fails 87% of its participants, the process itself is the product. No AI buildout timeline, and no national competitiveness argument built on one, survives contact with a five-year median wait.

Data centers turned a generation problem into a load problem

For most of the past decade, the queue crisis was a story about supply: solar farms, wind projects, and batteries waiting to connect. What changed in the past two years is the direction of the pressure. The crisis has flipped to the demand side, and the flip is most visible in Texas.

The Electric Reliability Council of Texas is tracking more than 438 GW of large-load interconnection requests, and nearly 90% of them come from data centers. To put that number in perspective, it is several multiples of ERCOT's current peak load. The momentum is accelerating rather than stabilizing: 198 GW of large load applied for interconnection in ERCOT in the first quarter of 2026 alone.

The demand-side flip matters because grid operators have decades of machinery for studying new generation and almost none for studying new load at this scale. A 500 MW data center campus stresses the system differently than a 500 MW power plant: it consumes rather than contributes, it wants firm service on an aggressive schedule, and it concentrates in clusters that strain local transmission. Texas responded first; the Public Utility Commission of Texas approved rules on June 18, 2026 for ERCOT to begin reviewing an initial set of large-load interconnection requests, known as Batch Zero. It is telling that the state regulator and the federal one acted on the same day. Both had reached the same conclusion: the existing rulebook simply does not contemplate the customer now knocking on the door.

The cost of delay is already priced in

The bottleneck is often framed as a future risk: capacity that will not arrive, growth that will not happen. That framing is too generous. Consumers are paying for the queue today, and the bill is visible in the one market mechanism designed to reveal scarcity: the capacity auction.

The trajectory in PJM, the largest US grid operator, is stark. PJM capacity auction prices rose from about $29 per MW-day for the 2024-25 delivery year to about $270 per MW-day for 2025-26, and prices for the 2026-27 delivery year hit the FERC cap of $329 per MW-day across the entire PJM footprint. That is an eleven-fold increase in two auction cycles, halted only by a regulatory ceiling. Prices did not rise because generators became more expensive to run; they rose because new supply could not get through the queue while new demand kept arriving.

The counterfactual sharpens the point. Analysis commissioned by GridLab and executed by Aurora Energy Research concluded that if just 10% of the 107 GW of land-based renewables in the queue before 2024 had been built in time for the 2026-27 auction, it would have added 1.5 GW of net supply, with material savings for ratepayers. The queue, in other words, is not a neutral waiting line; it is an active transfer mechanism, converting administrative delay into higher electricity bills. The failure to connect new, cheaper generation in the PJM market cost consumers an estimated $7 billion in a single capacity auction. Every month the reform saves is a month of that transfer avoided.

FERC chose speed over uniformity

Against that backdrop, the design of the June orders makes sense. FERC faced a choice between two reform architectures. The first was a comprehensive national rulemaking: uniform, durable, and slow. FERC had opened a rulemaking docket that accumulated an administrative record of over 3,500 pages of comments from nearly 175 industry stakeholders, the standard raw material for a multi-year proceeding. The second was targeted enforcement: narrower, regionally fragmented, and fast. FERC chose the second.

The mechanics are aggressive by regulatory standards. Each RTO has 60 days from June 18, 2026 to demonstrate that its existing tariff remains just and reasonable or to submit revisions addressing FERC's concerns, and grid operators and transmission owners must submit informational reports within 30 days describing how they plan to ensure sufficient generation resources are available to serve both existing and new large-load customers. Requests to pause are actively discouraged: FERC stated that it will heavily scrutinize abeyance requests, that any abeyance will be limited to 90 days, and that it will disfavor requests to extend the abeyance period.

Note the second requirement carefully. FERC is not only demanding faster interconnection process; it is demanding that grid operators show their homework on generation adequacy. That is the Commission implicitly acknowledging the fast-lane-to-nowhere problem: a shortened queue is worthless if there is nothing behind it. The regional approach also has a defensive logic. By deploying targeted, region-specific orders restricted to FERC's jurisdiction rather than a broad national rulemaking, FERC maximizes the legal durability of its actions against potential challenges from state-level stakeholders. Speed was purchased with fragmentation; six dockets will now evolve six answers.

The unresolved fight is who pays

Timelines can be ordered. Cost allocation must be negotiated, and it is here that the reform will be won or lost.

The federal position has been unusually blunt. The orders trace back to a 2025 request submitted by Department of Energy Secretary Chris Wright aimed at accelerating grid interconnection for large-load energy users, and the DOE framework did not dodge the distributional question: it recommended standardizing study deposits, studying load concurrently with generation, and assigning 100% of network upgrade costs to the interconnecting load.

That last principle is the fault line. If data centers bear the full cost of the transmission upgrades their connections require, the reform is politically sustainable; ratepayers are shielded, and the AI industry pays for its own on-ramp. If cost responsibility blurs, as it historically has in transmission planning, the PJM auction dynamic repeats in a new form: the public funds private infrastructure through opaque channels, and the backlash eventually reaches the reform itself. The six RTO compliance filings due this summer will reveal where each region lands, and the variance between them will define the investment map for the next phase of the buildout. Watch the cost allocation provisions, not the timeline provisions; the timelines are the headline, but the money is the story.

Conclusion

FERC's June orders deserve the significance attached to them. They compress a decade of drift into a 60-day clock, they force grid operators to answer for generation adequacy rather than just process speed, and they arrive while the cost of inaction is compounding in capacity auctions. As a piece of regulatory engineering, the show cause architecture is shrewd: fast, legally defensible, and regionally adaptive.

But the honest verdict has to be conditional. The orders fix the queue, not the grid. The 2,000 GW backlog, the 13% historical completion rate, and the $329/MW-day price cap all describe a system whose deeper constraints are physical: transmission that takes a decade to build, turbines with multi-year lead times, and a generation fleet that retires faster than it is replaced. Process reform was necessary, and FERC has now delivered the most serious version of it in years. Whether it proves sufficient will be decided not in the six dockets, but in what gets built behind them. The paperwork is now moving at AI speed. The electrons are not, yet.