Unpacking DOE’s proposal to transform data center interconnection

Procedurally this is a letter, not a rule. Secretary Chris Wright used a provision known as 403(b) in the statute that created the Department of Energy to direct the Federal Energy Regulatory Commission to consider issuing an advance notice of proposed rulemaking on large-load interconnection. It runs fourteen pages against the twelve hundred of the regional transmission planning rule the commission issued the previous year, so it is conceptual by design. The commission can decline but must justify declining; the precedent people remember is the 2017 letter under the neighboring provision asking it to subsidize coal and nuclear plants that kept fuel on site, which it rejected unanimously. Reception this time looks different. Rosner signaled eagerness, Senator Mike Lee supported it from the other side of the aisle, and the stakeholders Norris has seen view it favorably, though he adds that perspectives not yet heard will surface and that the jurisdictional piece will draw objections from state commissions and investor-owned utilities.

That jurisdictional claim is the part most likely to be fought. Under the Federal Power Act the commission has the transmission system and wholesale sales, while states keep generation and distribution. The fuzzy area is the transmission portion of a bundled retail bill in a state with a vertically integrated utility, where the state commission has historically overseen everything, including when a new load may hook up. The commission has never asserted authority there, which Clements attributes to tradition or practice rather than law: it does have jurisdiction over practices affecting transmission rates, and little affects those rates more directly than new loads connecting and the costs they impose. Nothing about who does the work would change. The utility still runs the study, still connects you, still names the cost, and the retail sale from a generator to a data center stays with the state. What changes is who supervises, which is what makes a standardized national approach possible instead of the current patchwork. Norris frames the gap as an old puzzle: generator interconnection was standardized twenty-three years ago and loads never got the equivalent.

The substantive heart is less about speed than about an accounting error. Today a load interconnection request and a generation interconnection request are considered independently, so when a utility studies a data center’s withdrawal during the most stressed hours to decide what upgrades the grid needs, it ignores an on-site generator or battery that would be offsetting that withdrawal during exactly those hours. The project therefore looks far more likely to trigger major network upgrades, which cost a great deal and take years. Study the two together and in many cases that requirement shrinks, which is why battery deals sized to the full nameplate of a facility have started appearing, one announced as accelerating interconnection by years. The curtailment fast lane applies the same logic explicitly: a load that will stop drawing when the grid operator asks does not impose the same worst-case burden, so it should not wait in the same line.

The catch is that almost everything determining whether this works sits in details the letter does not contain. Clements’ read, offered as instinct, is that the commission rarely regulates to that level of specificity, has been skittish about mandating standardization since standard market design collapsed in 2000, and issues principles instead. Layer on a timeline where an advance notice draws comments, becomes a proposed rule, draws comments again, and only then becomes a rule. Her worry is a rushed rule that never defines what a curtailment service is, whether you are selling energy or capacity, for how long, and how you get paid. Norris supplies the shape of the missing answer: bounded flexibility, a defined maximum of curtailed hours per year plus limits on event duration, since unbounded obligation is exactly what large loads refuse. Partial precedent exists in conditional firm transmission service, and the UK runs a curtailable connections program that compensates the customer when curtailment exceeds the guarantee. Both guests also warn against a rule so tied to on-site equipment that it disadvantages nearby but not-behind-the-meter options, with the Federal Power Act’s non-discrimination requirement as the guardrail. And Norris makes the argument that softens the whole risk: even if no rule issues, the record the proceeding generates would be portable to states and grid operators.