NRC Part 57 and Autonomous Operation, Explained

Where the proposed rule actually stands today

The Nuclear Regulatory Commission published proposed 10 CFR Part 57 in the Federal Register on May 1, 2026, as a licensing framework for microreactors and other reactors with comparable risk profiles ^[1]. The public comment period closed on June 15, 2026. The rule is not final.

Three consequences follow, and they are worth stating before anything else because most coverage blurs them. No developer is licensed under Part 57, because a proposed rule confers no licensing authority. The text can change between proposal and final rule, sometimes substantially. And a developer describing itself as operating under Part 57 today is describing an intention rather than a status.

Companion draft guidance, NUREG-2271, was issued for comment alongside it, framed by the NRC around rapid licensing of first-of-a-kind microreactors and high-volume deployment ^[2]. Guidance and rule move together, and both are drafts.

None of that makes the proposal unimportant. It makes it a direction rather than a destination, and reading it as a direction is what lets you plan against it honestly.

The problem Part 57 is trying to solve

The NRC has been circling microreactor policy for years. SECY-20-0093, in October 2020, flagged autonomous operation, remote operation, staffing, and regulatory oversight as open policy questions specific to this class of reactor ^[3]. Those questions did not have clean answers inside a framework built for large light-water plants.

More recently the agency has been planning explicitly for repetition. SECY-25-0052 addresses nth-of-a-kind microreactor licensing and deployment, including standardization of operational programs ^[4], which is the regulatory shape of many identical units rather than a handful of bespoke ones.

The statutory push comes from the ADVANCE Act, which directs the NRC to develop microreactor strategies across eight areas including staffing and operations, and oversight and inspections ^[5]. Congress asked for the thing Part 57 is attempting.

Underneath all of it is an arithmetic problem. A licensing and oversight model that assumes a large staffed plant does not scale to many small ones. If each microreactor consumes the regulatory attention of a conventional unit, the deployment numbers the industry describes are not reachable. Part 57 is the NRC trying to change that ratio deliberately rather than letting it be eroded by pressure.

What the rule changes about who has to be present

To see what is actually being proposed, you have to look at the requirement it sits against. Under 10 CFR 50.54(m), a licensed senior operator must be in the control room at all times, and a licensed operator or senior operator must be present at the controls at all times ^[6]. That is a condition of the license for the operating fleet, not a convention.

Proposed Part 57 contemplates remote operation and reduced on-site staffing for microreactors ^[1]. That is the substantive shift: not automation for its own sake, but relocating and reducing the human presence that current regulation fixes in place.

A distinction the NRC and INL have drawn matters here, because the industry uses these words loosely. Remote means command and control moved outside the reactor site boundary, and monitoring, meaning collecting and observing plant data, is a different activity from operations ^[7]. A vendor claiming remote capability may mean either, and the two carry very different regulatory weight.

Sandia National Laboratories, working for the NRC, described the operational picture the rules would have to accommodate: operators may not be located on site and may monitor from a remote location, and some designs contemplate one control room supervising multiple microreactors ^[8]. That last clause is the fleet-scale question, and it is the one with the least settled answer.

It is worth being concrete about why this could not simply be handled by exemption. Oak Ridge National Laboratory examined what autonomous control actually disturbs in the existing regulatory structure and found the list runs well beyond headcount: staffing requirements, manipulation of the controls, licensed operator provisions, technical specifications, cybersecurity, and event notification obligations, with the added wrinkle that a control room may not be co-located with the plant at all ^[18].

That is the case for a purpose-built framework rather than a series of carve-outs. Each item on that list is a separate place where regulation written for a staffed plant assumes a person who is now somewhere else, or nowhere. Granting exemptions one at a time would leave a licensing basis stitched together from exceptions, which is difficult to review consistently and harder still to replicate across many identical units. Part 57 is an attempt to write the assumptions down once.

The tradeoff is that a new framework has to earn its own confidence. An exemption sits against decades of operating experience with the underlying rule. A new rule for a class of reactor that has not yet operated commercially has no such record behind it, which is part of why the oversight provisions discussed further down matter as much as the licensing ones.

Part 57 compared with the final Part 53 rule

Two frameworks are often mentioned together and they are not interchangeable. Part 53 is final. Part 57 is proposed and microreactor-specific.

Part 53 was finalized as a risk-informed, technology-inclusive framework for advanced reactors ^[9]. It is optional, and it is available now. Part 57 is the narrower, faster instrument aimed at a specific class, and it is not.

The practical read for a buyer: ask which framework a developer is pursuing and why. A developer betting entirely on a rule that has not been finalized is carrying a schedule risk that a developer using an available pathway is not. Neither choice is wrong. The absence of an answer is the signal.

What autonomous operation does not mean

The proposed rule introduces vocabulary, and vocabulary is where most of the public confusion lives. Three terms get used interchangeably and should not be.

Automation is a machine performing a defined function without a person executing it, and it has existed in reactors for decades. Remote operation is command and control from outside the site boundary. Autonomous operation describes a system taking action across a range of conditions without an operator directing each one. A plant can be heavily automated with people on site, or lightly automated and operated remotely.

What none of them means is a reactor with nobody responsible for it. Safety-significant actions keep a human in the loop, and no facility today is licensed to operate unattended. "Unmanned" and "fully autonomous" are not regulatory categories, which is why we treat them as language to avoid rather than goals to advertise, and why we cover the terminology distinctions on their own.

The framing that has held up best comes from Brookhaven National Laboratory, in work for the NRC on facilities without main control rooms: the safety question is not so much justifying why a design has no main control room, but rather verifying that important human actions can be accurately and reliably performed ^[10]. Autonomy does not remove the human actions. It changes how anyone confirms they happened.

The oversight half of the rule that gets less attention

Licensing is only one side. The other is what happens for decades afterwards, and here the NRC has been unusually direct. In December 2025, staff proposed operational-phase microreactor oversight built on innovative inspection methodologies and a scalable inspection footprint ^[5].

Set that against the current baseline. The NRC keeps roughly 150 resident inspectors in the field, at least two at every plant, describing their role as independently verifying that requirements are being met ^[11], inside a Reactor Oversight Process built on inspection findings, performance indicators, and a significance determination process ^[12].

A scalable footprint across many small units means fewer inspector-hours per reactor. That is the explicit intent, not an unintended consequence. Meanwhile the Government Accountability Office has reported that the NRC has not evaluated its efforts to address staffing gaps and lacks benchmarks for whether recruitment and retention are working ^[13], and still lists licensing advanced reactors among its priority open recommendations ^[14].

So the trajectory is fewer regulator eyes per reactor, arriving alongside more reactors. Something has to carry the confirmation load that presence used to carry, which is precisely the verification problem the rest of this site is about.

There is a reasonable counterargument worth stating fairly. A microreactor is a smaller source term than a gigawatt-class plant, so proportionally lighter oversight is not obviously wrong, and the entire premise of risk-informed regulation is that attention should follow consequence rather than be distributed evenly. On that reading, a scalable inspection footprint is the framework working as designed rather than a weakening of it.

The response is not that the reasoning is wrong, it is that it is incomplete. Reduced consequence justifies reduced inspection intensity. It does not by itself establish how anyone confirms a reactor is behaving as described between those less frequent inspections. Those are separate questions, and the second is the one with no settled answer yet. A framework can be correct about proportionality and still leave a gap in confirmation, which is what we think is happening here.

The trust surface the proposed rule opens

If a reactor is operated from outside the site boundary and inspected less often, then more of what anyone knows about it arrives as data the operating organization produces about itself. That is not a criticism of any operator. It is a structural description of the model Part 57 contemplates.

It creates a question the rule does not answer, and arguably should not: who confirms the reported state, and are they separate from the party reporting it. Nuclear already contains the precedent for the answer, since IAEA safeguards exist so that an outside body can independently verify rather than rely on an operator's assertion ^[15]. Computing standardized the same split, with a verifier appraising evidence separately from the attester that produced it ^[16].

This is why self-attestation and independent verification are worth separating carefully when reading any vendor claim made under a Part 57 framing. The rule opens the operating model. It does not supply the trust layer that model needs, and no developer should imply that it does.

What this means if you are evaluating a developer, including us

RankShield Energy is a pre-applicant with the NRC. We hold no license, permit, or design approval, we are not licensed under Part 57 or any other rule, and nothing about our design has been demonstrated to or accepted by the NRC ^[17]. Describing the rule is not the same as satisfying it, and we are not claiming to.

Three questions travel well here. Which framework is the developer pursuing, and is it final or proposed. When they say remote or autonomous, which of the three definitions do they mean. And who confirms reactor state independently of the operator. Those are the questions in our vendor evaluation guide, and they are answerable without any proprietary disclosure.

Our own view, stated plainly so it can be argued with: the staffing and oversight changes Part 57 contemplates are reasonable, and the verification layer they imply is not yet built by anyone, including us. Treating that gap as solved is the most common overclaim in this market right now.