Part 57: Right-Sizing Nuclear Regulation for Low-Consequence Reactors

The NRC’s proposed new licensing framework, called Part 57, tests whether nuclear regulation can do something it often says it wants to do: match regulatory burden to actual risk.

Part 57 would create a new licensing framework for a narrow class of reactors, primarily microreactors and other designs with very low potential offsite consequences. If the rule works as intended, it would make it easier to license, manufacture, and deploy those reactors without weakening the NRC’s obligation to protect public health and safety.

Part 57 differs from existing NRC rules—such as Part 50 and 52 that are aimed at licensing larger reactors—because it works as a pathway for reactors that first satisfy restrictive entry criteria. The rule is built around the idea that some reactors are small enough, simple enough, and low-consequence enough that applying the licensing structure developed for large commercial power reactors may add cost and delay without adding commensurate safety value. The rule extends from decades of operations and oversight of university and research reactors.

But this is still a proposed rule. The entry criteria, emergency planning provisions, manufacturing and construction pathways, and assumptions behind remote and autonomous operation will determine whether Part 57 actually achieves risk-informed regulation—or simply creates a new set of boundary problems.

The Problem Part 57 Is Trying to Solve

The legacy commercial nuclear licensing system was built around large light-water reactors. Those plants have large source terms, complex safety systems, significant site-specific infrastructure, and potential accident consequences that justify extensive review. The regulatory architecture reflects that history.

The new Part 53 rule, effective as of April 29th, establishes a technology-inclusive licensing framework that is better aligned with advanced reactor technologies and modern risk analysis. The framework reflects a transition away from purely deterministic regulation, but not fully into a risk-informed, performance-based system. Microreactors and some other advanced designs do not fit neatly into either of those models. They may have far smaller inventories of nuclear material, lower decay heat, simpler safety cases, and less dependence on operator action. Some may be manufactured in factories, shipped to deployment sites, monitored remotely, and used for applications that are not conventional grid-scale electricity generation.

If a reactor has a small source term, low decay heat, passive or inherent safety characteristics, and accident consequences bounded below the proposed dose threshold, then many traditional requirements may become poorly matched to the risk. Under those conditions, repeating large-reactor procedures can result in cost and delay without corresponding safety gains.

This is especially important for microreactors because their proposed value depends on deployment models that current licensing processes may not accommodate well. If each deployment is treated as a bespoke large-reactor project, many use cases may never become commercially viable, despite more than reaching safety standards. That may be acceptable if the risks justify the burden.

Part 57, and all risk-informed regulation, starts from the premise that regulation should be proportionate to consequence. A framework that treats fundamentally different risk profiles as if they were the same is not more protective by default. It can also be slower, less predictable, and less focused on the safety issues that matter most.

A Screened Pathway, Not a Universal Shortcut

The most important feature of Part 57 is that it is not available to every reactor. It is a screened pathway. A reactor must first meet entry criteria intended to bound its consequences before it can use the streamlined provisions.

The proposed rule uses two primary gates.

The first is a dose-based criterion. The applicant must show, with reasonable assurance, that an individual in the unrestricted area following a bounding postulated accident would not receive more than 1 rem total effective dose equivalent for the duration of the accident.

The second is a fuel-mass limit. The total inventory of thorium, uranium, and plutonium must not exceed 10 metric tons.

These criteria define the scope of the rule. If a design cannot meet them, it must use another licensing framework. If a design can meet them, the NRC is proposing that some traditional requirements can be simplified or removed because the potential public consequences are already tightly bounded.

That structure is coherent. It is also where much of the scrutiny should focus.

Historically, nuclear regulation has often used reactor capacity as a rough proxy for hazard. That approach is imperfect. Electricity output and thermal energy are not, by themselves, the same thing as safety risk. New reactor designs are increasingly separating electricity generation from the safety case. In some designs—such as the TerraPower Natrium design—the power conversion system is not part of the licensed safety system. In microreactor applications, some deployments may not involve electricity generation at all.

Fuel mass may be a better proxy than power in some contexts because it relates more directly to source term and residual heat. A smaller inventory can limit the amount of radioactive material available for release and reduce the decay heat that must be managed after shutdown. It may also reduce some security concerns.

But fuel mass is still a proxy. A 10 metric ton limit may be administratively clear and efficient, but it is not the same as a performance-based criterion tied directly to decay heat removal, accident consequences, or site-boundary dose. The NRC appears to recognize this issue by asking for input in the proposed rule on whether a performance-based alternative would be preferable.

If the goal is to regulate based on consequence, then the rule should be clear about whether the fuel-mass limit is a technically justified screen, a conservative administrative boundary, or an indirect way to limit eligibility to a narrow subset of reactors.

Major Components of the Rule: What’s Similar and What’s Different

For reactors that qualify, Part 57 uses a licensing process that combines elements of existing frameworks but also differs from them.

Part 57 builds off of Part 53 in many ways. Part 50 uses a two-step model of a construction permit followed by an operating license. Part 52 allows an applicant to seek a combined license, resolving construction and operating authority in a single licensing proceeding based on a substantially complete design. Part 53 adopts the license types available in the other framework, but places them in a performance-based regulatory framework. Part 57 combines the two-step approach with the requirement for a final and complete design and analysis at the time of application.

That requirement functions as an additional practical entry condition, alongside the dose-based eligibility criterion and the fuel-mass limit. The difference is that Part 57 conditions its streamlined process on having the design essentially settled before review begins. That front-loaded posture is what allows the NRC to contemplate a single comprehensive safety review, potentially one adjudicatory hearing, and an operating license decision for a fleet of identical reactors within 6 to 12 months of application acceptance.

Part 57 also provides for standard design approvals and generic finality, allowing resolved design issues to be referenced in future applications. That matters for any technology intended to be replicated. If the NRC has already reviewed a design feature, operational program, or safety basis, repeated relitigation of the same issue should not be the default. Standardization only produces regulatory value if the licensing process actually recognizes standardization.

A notable departure is that Part 57 would use general licensing in a limited but important way. The NRC is not proposing a general license for an entire reactor. The agency concludes that a nuclear reactor remains a utilization facility requiring specific licensing controls under the Atomic Energy Act. Instead, it uses general licensing where the statute gives the Commission more room: important component parts, certain construction activities, and, separately, reactor operators at operator-independent facilities.

Second, Part 57 would create a more realistic path for factory-built reactors. Manufacturing licenses could allow reactors to be fabricated, fueled, and tested before shipment to a deployment site. This may be one of the most important provisions for microreactors because many of these designs depend on a deployment model that looks more like manufacturing than traditional site-specific nuclear construction.

Third, for construction, the proposed general license would apply only to limited activities for nth-of-a-kind facilities that reference an approved design with generic finality. It would not allow special nuclear material or the manufactured reactor to be brought to the site before the construction permit is issued. The provision is designed to accelerate site preparation and deployment while preserving the licensing boundary for the reactor itself.

This could accelerate deployment, but it also shifts project risk to the developer. If a company begins construction before receiving final approval, it assumes the risk that the project may need to change—or may not be approved. That risk allocation is not inherently inappropriate. Developers routinely make investment decisions before every regulatory uncertainty is resolved. But the rule should be clear about what issues are genuinely settled, what remains open, and what risks are being assumed by the applicant rather than transferred to the public.

Fourth, Part 57 would also allow more flexible approaches to quality assurance and financial qualification.

Emergency Planning Shows the Logic of the Rule

Emergency planning is where the proportionality argument becomes easiest to see.

Part 57 would still require emergency plans. It would require applicants to address accidental releases, loss of control of radioactive material, and related hazards. It would also require coordination with offsite organizations.

What it would not require is a predefined emergency planning zone (EPZ) for qualifying reactors.

For a large reactor, an emergency planning zone reflects the possibility that protective actions could be needed beyond the site boundary. For a reactor that must first satisfy a stringent dose-based entry criterion, the logic changes. These designs would, by meeting the Part 57 entry criteria, already exceed regulations that allow for a site-boundary EPZ in existing frameworks. If a design has already shown that a bounding accident would not exceed the proposed site-boundary dose threshold, repeatedly evaluating whether it needs a large predefined EPZ may become redundant.

That does not mean emergency planning becomes unnecessary. A site-specific emergency plan still needs to be credible, robust, and depend on confidence in the underlying accident analysis. Offsite coordination may still be needed. Emergency workers still need protection. But, the form of emergency planning should reflect the consequences the reactor can plausibly create.

This is not an unprecedented concept in nuclear regulation. Research and university reactors with limited hazards have often been regulated differently from large commercial reactors. Many have emergency plans without the same type of offsite emergency planning zones expected for large power reactors. Part 57 would extend a version of that proportionality logic into a commercial framework.

Remote Operation and Autonomy Need Institutional Scrutiny

The operational provisions deserve greater attention. For large commercial reactors, the regulatory system assumes a staffed control room, extensive site-specific operational oversight, and a resident inspection model. Those assumptions make sense for the larger plants the system was built to regulate.

Part 57 would introduce operational flexibility that goes beyond schedule reform. It would allow remote monitoring, remote operation, and autonomous functions. It would adjust staffing and training requirements, including generally licensed reactor operators for facilities where operator action is not central to maintaining safety, which was introduced in Part 53. For personnel, the generally licensed reactor operator concept reflects a different operational premise: if a facility demonstrates that operator action is not required to maintain the reactor within the relevant safety criteria, the regulatory focus can shift from individual operator licensing to the licensee’s training, qualification, and proficiency program. That is potentially sensible for operator-independent microreactors, but only if the underlying safety case is durable. Part 53 refers to self-reliant-mitigation vs. interaction-dependent-mitigation facilities. Part 57 uses the terms operator-independent and operator-dependent facilities.

For low-consequence microreactors, the safety case may be different. If no prompt operator action is needed to protect the public, then licensing should not mechanically require staffing models designed for plants where operator action is safety-significant. If a qualifying reactor cannot produce offsite consequences beyond tightly bounded limits, then a large-site operational model may not add meaningful protection.

But remote and autonomous operation also change the failure modes. They change how problems are detected, diagnosed, and corrected. They change the role of operators and inspectors. They may reduce some human-error pathways while creating others.

That means the right question for stakeholders to consider in their comments is not whether remote or autonomous operation should be allowed, it is what institutional controls, inspection models, response expectations, and data access requirements are needed when human presence is reduced.

If the NRC does not anticipate full-time resident inspectors at these facilities, it should be clear about what replaces that model and why it is adequate. If operator action is assumed not to be safety-significant, the rule should explain how that assumption is demonstrated and maintained over the life of the facility.

The NRC declined to include some of these flexibilities in Part 53, even though the same microreactor designs could apply under Part 53. Part 57, therefore, becomes an important test case. If these changes are justified by the low-consequence entry criteria, the NRC should make that logic explicit. If they are justified by technology characteristics that could exist outside Part 57, then similar logic may eventually apply elsewhere. Otherwise, the changes are not risk-informed, but arbitrarily applied to a single licensing framework, which is the exact situation Part 53 tried to remedy.

Siting, Environment, and Non-Traditional Deployment

The proposed rule includes environmental and siting provisions intended to support deployment in non-traditional locations. These include potential categorical exclusions under NEPA for qualifying projects and more flexible approaches to siting, allowing applications for large geographic areas rather than only a narrowly defined site in every instance.

Part 57 also provides a mechanism for “generic finality” of site parameters, like seismic and weather limits, reducing unnecessary repetition by considering those matters resolved for future applications, but they also compress opportunities for case-specific review. That tradeoff is more acceptable for nth-of-a-kind reactors with complete approved designs than for first-of-a-kind deployments. The rule should maintain that distinction.

That is relevant because many microreactor use cases are not conventional centralized power plants. Potential applications include remote communities, industrial facilities, defense installations, disaster response, mining operations, heat or power supply in locations where large grid-connected plants are impractical, and even marine applications.

A licensing framework that assumes every deployment looks like a large, site-specific power station may make those use cases impossible, regardless of whether the reactor is safe enough to support them.

This does not mean siting and environmental review should disappear. It means they should focus on the issues that actually vary by site and matter for the facility’s consequences. If a reactor is standardized and low-consequence, the review should not repeatedly reopen generic design questions under the guise of site review.

Part 57 Clarifies What Counts as a Reactor

Part 57 also addresses a separate but important jurisdictional issue: what counts as a nuclear reactor and, therefore, what must be regulated as a utilization facility under the Atomic Energy Act (AEA).

Some recent arguments, including an ongoing lawsuit, have suggested that the Atomic Energy Commission (AEC) and NRC never squarely considered whether certain small reactors should be treated as utilization facilities. The proposed Part 57 rule makes that argument harder to sustain. The NRC states that it considered whether some reactors otherwise eligible for Part 57 could be excluded from the definition of “utilization facility” and regulated under a different framework. It rejected that approach.

The NRC is not creating a new definition, but it is reaffirming and carrying forward a longstanding one: a nuclear reactor is an apparatus, other than an atomic weapon, designed or used to sustain nuclear fission in a self-supporting chain reaction. In plain terms, if something can achieve sustained criticality, the NRC treats it as a reactor.

What is new is how directly the agency ties that definition back to the AEA The NRC explains why that definition matters: a self-sustaining fission reactor uses special nuclear material in a way that implicates both common defense and security and public health and safety.

The agency also considered whether some microreactors or advanced reactors could be excluded from the AEA’s utilization facility framework and regulated some other way. It rejected that option, concluding that any reactor capable of sustained criticality falls within the statutory definition of a utilization facility and should remain subject to licensing regulations.

That does not eliminate every legal argument about NRC jurisdiction over small reactors. But it changes the posture of the debate. The agency has now directly considered the issue in a rulemaking record and explained the technical basis for treating these reactors as utilization facilities. Part 57 therefore, reinforces the boundary between streamlining regulation and avoiding reactor regulation altogether.

Digging Deeper on a Few Topics

Part 57 is directionally important, but several unresolved issues deserve careful attention.

The entry criteria may be over- or under-inclusive. The 10 metric ton fuel-mass limit is clear, but clarity is not the same as precision. The limit would include the NuScale power module and X-Energy XE-100, but exclude the GE Verona BWRX-300 and the TerraPower Natrium. As discussed earlier, if fuel mass is being used as a proxy for residual heat, source term, or security significance, the NRC should explain the technical basis for the threshold. If other designs can meet the same consequence limits with different fuel masses or larger physical configurations, the rule should explain why they should be excluded. Conversely, if the 10 metric ton limit captures hazards not fully reflected in the dose criterion, the NRC should make that rationale explicit.

But there is also a larger regulatory philosophy question. If physically larger reactors can demonstrate similarly low consequences, why should they not receive similarly risk-informed treatment at reactors that have less than 10 tons of material?

The relationship between Part 57 and Part 53 needs to be clarified. Part 53 is intended as a broader technology-inclusive framework. Part 57 is narrower and more streamlined. If Part 57 builds on Part 53 conceptually, the NRC should explain whether Part 57 is a special case within the same regulatory philosophy, or parallel framework, and why it couldn’t be regulatory guidance for Part 53 as a better version of the (ill-fated) AERI approach. The proposed rule’s treatment of reference information should also be internally consistent across Part 50, Part 52, Part 53, and relevant federal authorizations. Right now a Part 57 applicant can reference Part 50, 52, and authorizations from other agencies, but not Part 53 licenses.

A Promising Framework, Not Yet a Verdict

Debates over NRC reform often collapse into a familiar binary: the agency is either moving too slowly and blocking innovation, or moving too quickly and compromising safety.

That is the wrong frame for Part 57.

The better question is whether the NRC can distinguish among reactors that pose fundamentally different levels of risk and regulate them accordingly.

It has done so before. Research and university reactors have long been treated differently from large commercial power reactors because their hazards are different. Part 57 attempts to apply a similar principle to commercial reactors that meet stringent consequence-based criteria.

That is an important consideration of all of the NRC’s recent efforts to modernize. The NRC’s mandate is not to maximize regulatory burden. It is to provide reasonable assurance of adequate protection while enabling the benefits of nuclear technology where safety can be demonstrated. A system that imposes unnecessary burdens on very low-risk technologies can fail that mandate in a different way: by preventing beneficial technologies from being tested, improved, and deployed.

But the inverse is also true. A streamlined pathway is only justified if the entry criteria are meaningful, the oversight model is credible, and the operational assumptions are valid.

Part 57 could become a meaningful shift toward consequence-based regulation for a subset of nuclear technologies. It recognizes that not every reactor should be treated like a large commercial power plant. It creates a pathway for designs that may be manufactured, replicated, transported, remotely operated, or deployed in places where conventional licensing assumptions do not fit.

That is the right direction for reactors whose risks are genuinely bounded.

But the proposed rule is not self-validating. The NRC still needs to justify the entry criteria, clarify how deterministic limits relate to performance outcomes, explain the relationship with Part 53, and demonstrate that the proposed oversight model is adequate for remote, autonomous, and highly standardized deployments.

At this stage, the right position is neither blanket endorsement nor reflexive opposition for anyone. Part 57 deserves serious review because it asks the right institutional question: whether nuclear regulation can become more proportional without becoming less protective.

The answer depends on the details. Those details are now the work of the comment period, which is open for the next 45 days.