TerraPower wins NRC permit for Natrium reactor

The US Nuclear Regulatory Commission has approved Bill Gates-backed TerraPower to build a Natrium nuclear reactor in Wyoming, the first construction permit the NRC has issued in nearly a decade. TechCrunch reports the plant is planned for 345 megawatts and will be built near an aging coal facility, with TerraPower partnering with GE Vernova Hitachi. The approval is also notable because the design uses molten sodium coolant, the first non-water-cooled commercial reactor cleared by the NRC in more than 40 years.

The permit is a reminder that in nuclear power, capital is not the scarce input—permission is. TerraPower has raised about $1.7 billion, according to PitchBook via TechCrunch, and investors have poured more than $1 billion into nuclear startups recently amid rising electricity demand from data centers. Yet the bottleneck is the regulator’s calendar and the paperwork required to build on private land. The Department of Energy has loosened safety rules on DOE-owned sites, but those changes do not apply to projects that want to connect to the grid outside federal property. In other words, the fast track exists, but only if you build where the state is your landlord.

Natrium’s pitch is partly technical and partly commercial. Sodium cooling is presented as safer than water cooling, and the design includes large insulated tanks that store excess heat so the plant can keep running near full capacity while meeting variable demand. That feature is aimed directly at the intermittency problem created by wind and solar buildouts: instead of ramping the reactor down, the system stores heat and releases it later. The economic promise is lower generation costs by avoiding the inefficiencies of frequent power swings.

But the approval also highlights how industrial policy now runs through gatekeeping rather than direct subsidies. When an agency issues the first permit in almost ten years, it is not simply validating one company’s engineering; it is deciding who gets to compete at all. That decision shapes supply chains—who signs long-lead contracts, who hires, who can credibly raise the next funding round—long before a single kilowatt-hour is produced. It also creates a two-tier market: firms with the regulatory stamina and political access to survive a decade-long process, and firms that cannot afford to wait.

The current wave of nuclear enthusiasm is being sold as a response to AI-driven electricity demand. Yet the same dynamic that constrains grid buildouts—permits, interconnection queues, and public-process veto points—also governs nuclear. Mass manufacturing may reduce costs over time, as startups argue, but the savings arrive only after years of repetition, and repetition requires a pipeline of approvals.

TerraPower now has permission to build. The next question is whether it can build on the timeline implied by the permit, or on the timeline implied by the last decade.