NRC clears X-energy in 52-year first. Pennsylvania resets the grid cost default. #Nuclear #XEnergy

Added: 2026-05-25

[00:00:00]The Pennsylvania Public Utility Commission released a model tariff framework last week. It applies to large load customers, anything over 50 megawatts individually, 100 megawatts in aggregate. Translation, data centers and big new manufacturing. The central provision here, utilities should charge these customers up front for any grid infrastructure that would not have been needed but for their connection to the system. Cost causation internalized to the customer who triggered the cost.

[00:00:27]Now, that sounds like an accounting nuance, but it's actually a structural inversion. The default assumption for decades has been that the grid expansion costs get socialized across the ratepayer base because the existing customers were the default beneficiaries of any new infrastructure. The new default in Pennsylvania, though, is that the customer triggering the expansion absorbs it up front through what is called contributions in aid of construction. Pennsylvania may be the first state to require this for data center driven expansion. Now, on the same week, a different agency, the Nuclear Regulatory Commission, completed its environmental review of X-Energy's four-reactor project at Dow's Seadrift chemical complex on the Texas Gulf Coast and issued a finding of no significant impact. In 52 years of regulating commercial nuclear power, the NRC has never before cleared a project through an environmental assessment instead of the heavier environmental impact statement. The XE-100 design is a high-temperature gas-cooled reactor, a class of technology that the United States essentially abandoned in 1989 when Colorado's Fort Saint Vrain plant retired. What both moves here share, institutions of the prior era reaching for proportionality. Pennsylvania calibrating who pays for what, the NRC calibrating how much review fits the actual footprint. The growth in electrical load this country is about to absorb has no comparable precedent.