The Nuclear Co-Location Precedent and the Regulator's Answer

The abstract tension became a concrete national fight when a large data-center developer arranged to co-locate a very large load directly at a nuclear plant, drawing power straight from the reactor under a long-term deal, with the load scaling into the hundreds of megawatts. The plant owner and the load celebrated it as a template: fast, clean, baseload power for AI without waiting in the transmission queue. To make it work at full scale, the parties filed to amend the plant's interconnection agreement with the regional grid operator, because the arrangement changed how the plant's output would be used and how the combined facility would relate to the network. That filing put the deal in front of the federal regulator, and it turned a private commercial arrangement into a public test case.

The regional grid operator's stakeholders and other utilities objected, and their argument was the one from the previous lesson made specific. If a large load parks itself at a plant and still relies on the grid for backup and for stability, but pays none of the transmission charges that fund that grid, then its neighbors are quietly subsidizing it. The saved charges do not vanish; they shift onto everyone else's bills. Worse, taking a large baseload plant's output off the market to serve one private customer could tighten regional supply and raise prices for the remaining customers, and it could complicate the reliability the grid operator is obligated to maintain. The objectors were not opposing co-location in principle. They were insisting that a load leaning on the shared network must pay for the network like everyone else.

The federal regulator's response, in late 2024, was to reject the amended interconnection agreement as filed. Crucially, it did not rule that co-location is unlawful or impossible. It ruled that the specific arrangement, and the record supporting it, had not adequately resolved the questions the objectors raised: how the co-located load would be treated for transmission cost allocation, how its reliance on the grid for backup would be accounted for, and how reliability and cost-shifting concerns would be addressed. The regulator sent the parties back to build a proper record and, in effect, told the whole industry that the behind-the-meter label does not by itself answer the hard questions. If you want to co-locate at scale, you must show your work on who pays for the grid services you continue to use.

What followed was a regulator-convened proceeding to develop rules for large loads co-locating at generation, an acknowledgment that this is not a one-off but a category that needs a durable framework. The questions on the table are exactly the ones the physics forced: When is a load truly islanded and therefore fairly exempt from transmission charges? When is it grid-reliant and therefore obligated to pay? How should backup service from the network be priced? How do you protect other ratepayers and regional reliability while still allowing an arrangement that can bring new load and new investment quickly? The answers are still being written as I teach this, but the direction is clear. Co-location will be permitted, but on terms that make the co-located load pay for whatever grid services it actually uses.

For an operator or investor the precedent carries a plain lesson. Do not underwrite a co-location deal on the assumption that the behind-the-meter structure makes the interconnection questions disappear. Underwrite it on the assumption that a regulator will eventually ask how your load relies on the shared grid and will require you to pay for that reliance. The deals that will survive are the ones designed from the start to answer that question honestly, whether by genuinely islanding the load, by contracting for and paying for explicit backup service, or by accepting curtailment when leaning on the plant alone would strand the wider system. The nuclear case did not close the door on co-location. It set the price of admission, and the price is paying your own way onto a grid you still quietly depend on.