Arturo Espinosa – For most of the twentieth century, nuclear power meant one thing: a large light water reactor—a massive plant with iconic cooling towers, where ordinary water is used to cool and regulate a uranium-fueled reaction that produces electricity. For decades, the technology was settled and had a well-understood risk profile. The regulatory architecture built around it, codified in 10 C.F.R. Parts 50 and 52, reflected that certainty. These regulations prescribed in exacting detail how a reactor had to be designed, constructed, and licensed, leaving little room for variation. They assumed a stable technological baseline and, in doing so, effectively locked the regulatory system to a single reactor paradigm.
That assumption no longer holds.
Over the past decade, a generation of developers has brought forward new nuclear technologies sharing almost nothing with the conventional light water reactor. Unlike the large, high-pressure, water-cooled reactors used for decades, these newer designs run at lower pressures, use alternative coolants such as molten salt or helium, and often eliminate the risk of meltdown or dramatically reduce long-lived waste. Small modular reactors, molten salt designs, gas-cooled systems, and liquid metal-cooled fast reactors, for example, each carry a different risk profile, different failure points, and different safety characteristics. And then there is fusion: not a variant of fission at all, but a fundamentally different process. Fusion forces light atomic nuclei together rather than splitting heavy ones, releases energy without a self-sustaining chain reaction, cannot melt down, and produces none of the long-lived radioactive waste associated with traditional fission reactors.
The traditional regulatory framework under Parts 50 and 52 could not accommodate these new developments. The Part 50 and 52 framework imposes detailed, design-specific requirements—dictating particular cooling systems, fuel compositions, and containment structures—built around traditional reactor design. There are many ways to build a safe reactor, but regulators were stuck measuring all of them against an outdated template. Applying those rules to reactors that do not use pressurized water, do not rely on the same fuel, and cannot fail in the same ways produced absurd results. Developers were forced to demonstrate that they had mitigated hazards their technology does not even create.
Congress recognized this gap in passing the Nuclear Energy Innovation and Modernization Act, directing the NRC to build a technology-inclusive framework for advanced reactors. The result was the recently published 10 C.F.R. Part 53 (released March 30, 2026)—a performance-based system that defines safety objectives and measurable outcomes, allowing developers to meet those benchmarks through different design and engineering approaches rather than prescribing a single method. For fission-based advanced reactors, Part 53 delivered what the industry needed.
Fusion, however, was left out entirely. And the omission was not accidental.
The exclusion reflects a classification problem the NRC has never fully resolved. The Atomic Energy Act was written when the relevant technologies were fission reactors and radioactive materials; commercial fusion simply was not part of the legislative imagination. The agency has since spent years debating three possible approaches.
First, to regulate fusion as a “utilization facility“—the same category governing traditional fission reactors—subjecting fusion developers to the full weight of reactor licensing. The legal basis is plausible because Part 50 defines a utilization facility broadly enough that any installation producing usable atomic energy could arguably fall within its scope, but the practical consequence would be prohibitive: licensing under this framework costs hundreds of millions of dollars and takes years—a burden threatening to end most fusion programs at their current nascent stage.
Second, to treat fusion as a byproduct material under Parts 30 through 37. The connection is tritium, the hydrogen isotope most fusion devices use as fuel. Because tritium is classified as a byproduct material under the Atomic Energy Act, fusion facilities can be brought under the Part 30 framework without new statutory authority. The approach offers near-term clarity at a fraction of the cost of reactor licensing.
Third, a hybrid: requirements calibrated to actual risk, with reduced oversight for early-stage systems and stricter requirements triggered as facilities scale. The hybrid approach most closely mirrors the principle underlying Part 53. However, it would require new rulemaking the NRC has not pursued.
In 2023, the Commission chose the second option.
Choosing the Part 30 byproduct framework has enabled real progress. The Part 30 framework imposes necessary safety requirements without prohibitive cost. For today’s fusion experiments, this calibration seems appropriate because current devices are still at the demonstration stage, where the risks are minimal and well within what Part 30 was built to handle.
But the arrangement carries structural vulnerabilities. Because Part 30 authority can be delegated to the states, states have moved to fill the regulatory vacuum. Some of the most consequential decisions about where and how fusion develops are now being made at the state level. Virginia and Tennessee, for example, have adopted accommodating postures. Virginia, where Commonwealth Fusion Systems plans to build its first grid-scale ARC plant, and Tennessee, where Type One Energy is deploying its stellarator prototype at a retired TVA coal site, have both worked closely with developers to align licensing with project requirements. However, the risks of fragmentation are equally real. Massachusetts, for example, requires developers to get a state license and demonstrate rigorous procedures to safely control and use radioactive materials. As such, developers face inconsistent requirements across jurisdictions, and the current approach rests on an uncertain legal foundation meant to serve as a placeholder until the issue can be more fully considered.
The solution is a hybrid framework that applies regulatory intensity proportional to actual hazard, with clear thresholds as systems scale toward commercial generation. Such an approach is what Part 53 achieved for fission. Fusion deserves the same treatment—a framework designed for what it actually is, not a provisional classification borrowed from medical isotope labs. Until a hybrid framework exists, fusion develops in a gray zone: lightly regulated, jurisdictionally fragmented, and shaped more by state competition than federal coherence. While the Part 30 byproduct framework may sustain innovation in the short term, it will not sustain an industry in the future.
