How the Nuclear Regulatory Commission Works — And Why Everyone Has an Opinion About It
The agency that decides whether America's nuclear future happens is itself in the middle of the biggest overhaul in its 50-year history.
The Nuclear Regulatory Commission is probably the most consequential agency you’ve never thought much about. It decides whether new reactors get built. It sets the safety rules those reactors must follow. It approves the designs, the sites, the fuel, the emergency planning zones, the operator training programs. If an SMR developer wants to put a reactor online in the United States, every path runs through the NRC. Every single one.
That makes the NRC enormously powerful, and right now, it’s under more pressure than at any point since it was created in 1974. The nuclear industry says it’s too slow. Safety advocates say any speed-up is reckless. The Trump administration ordered a “wholesale revision” of its structure and culture. Congress passed bipartisan reform legislation. A new regulatory framework for advanced reactors just took effect in March 2026. And career NRC staff are — according to multiple recent reports — genuinely worried about whether the agency can maintain its independence. If you care about whether SMRs actually get built in America, and built safely, you need to understand how this agency works and why the debate around it is so heated right now.
From atomic watchdog to nuclear gatekeeper
The NRC didn’t always exist. Before 1974, nuclear power in the United States was regulated by the Atomic Energy Commission, a body that combined both the promotion and the oversight of nuclear technology in a single organization. You might think that sounds like a conflict of interest, and you’d be right. Congress thought so too, which is why it passed the Energy Reorganization Act of 1974 and split the AEC in two: the Energy Research and Development Administration (which eventually became the Department of Energy) handled promotion and development, while the newly minted NRC took over safety and licensing. 🏛️
Under the AEC’s watch, the United States had built 54 reactors that were operating by 1974, with 70 more under construction or licensed. Costs in 2025 dollars ran under $3,000 per kilowatt, and the median construction time was about six years. That sounds almost impossible compared to modern timelines, and the nuclear industry never tires of pointing it out.
What happened next is a genuinely contested historical question. After the NRC took over, new regulations introduced in the 1970s and early 1980s tripled construction costs and timelines, according to an analysis in The National Interest, resulting in the cancellation of most projects then in progress. The reasons include:
Post-Three Mile Island regulatory changes applied retroactively to plants already under construction
An adjudicatory hearing process that allowed opponents to delay projects for years
Design-change requirements issued during construction, forcing expensive rework
Per-hour fee charges to applicants, creating massive unpredictability in licensing costs
In the 50 years since the NRC was created, it has licensed exactly two reactors from start to finish — Vogtle Units 3 and 4 in Georgia — projects that required more than 14 years and ran billions of dollars over budget. The original fleet, licensed under the old AEC system, is still operating safely, with many reactors now approved for 80-year service. France, for comparison, built 55 of its 57 reactors in 15 years using standardized designs and consistent safety procedures. South Korea regularly completes large reactors in five years. The NRC’s defenders will say those are unfair comparisons. But the gap is hard to dismiss. ⚛️
How nuclear licensing actually works
Walk into the NRC’s official licensing process and you’ll find something that resembles a choose-your-own-adventure game designed by people who love regulatory certainty more than readable prose. There are currently two main pathways for licensing new reactors, under 10 CFR Part 50 and 10 CFR Part 52, and as of March 2026 there’s a new optional third pathway called Part 53. Each has its own rules, timelines, and application types.
The old Part 50 framework — the one most new reactor applications have historically used — grants construction permits but not operating licenses. Those come later, separately. The Part 52 framework, introduced in 1989, improved this by combining construction and operating authorization into a single Combined License (COL), but it’s still built entirely around light-water reactor technology. If you’re building a sodium-cooled fast reactor like TerraPower’s Natrium, or a molten salt reactor, or an air-cooled microreactor, you’re trying to fit an inherently new thing into a regulatory framework that wasn’t built for it. 🔧
The key steps in the current licensing process look roughly like this:
Early site permit (ESP): Secure site approval before committing to a specific design
Design certification: Get the reactor design itself reviewed and approved, independent of site
Combined license (COL): The merged construction and operating authorization
Mandatory hearing: Mandatory public adjudicatory process, even absent formal opposition
Inspections, tests, analyses: Verification work before actual fuel loading is allowed
Each stage involves staff review, public comment periods, possible contested hearings before the Atomic Safety and Licensing Board, and Commission votes. The Sidley Austin analysis of Part 53 notes that the new framework lists eight distinct types of license applications. Eight. That’s the streamlined version. The NRC also charges applicants by the hour for staff review time, which was running at $318 per hour until the ADVANCE Act cut the rate for advanced reactor applicants to $148 — a meaningful improvement, but still a billing model that turns regulatory uncertainty into a direct financial liability. 📋
Why critics say the NRC is broken — and where they disagree
The criticism of the NRC comes from two directions, and the two sides don’t agree on what exactly is wrong.
The first argument, dominant in the nuclear industry, goes roughly like this: the NRC was designed to regulate massive one-of-a-kind light-water reactors, using safety rules written in the 1970s, and it has never fundamentally updated its institutional architecture to deal with mass-produced, passively safe, modular designs. The Breakthrough Institute published a sharp analysis in February 2026 arguing that the NRC’s problem isn’t how cautious it is — it’s how it’s structured. The agency spent decades optimizing for a single objective — accident prevention — while being asked to govern a system where multiple objectives now matter simultaneously. The result, according to that analysis, is a regulator that became “extraordinarily good at answering one question — ‘Is this safe enough?’ — and institutionally incapable of answering another — ‘Does this regulatory choice serve the general welfare better than the alternatives?’” ⚠️
The second argument comes from safety advocates and some former NRC officials, and it goes in the opposite direction: the agency is under political pressure to move faster than is wise, and the independence it needs to make hard safety calls without political interference is genuinely at risk. Allison Macfarlane, who served as NRC chair during the Obama administration, put it plainly in a recent ProPublica investigation: “The regulator is no longer an independent regulator — we do not know whose interests it is serving. The safety culture is under threat.”
Both arguments contain real information. The NRC was built for a technology that barely resembles today’s SMR designs. And the NRC is under extraordinary external pressure right now. These facts don’t cancel each other out — they coexist uncomfortably, which might be why the debate around this agency is so hard to resolve. What’s your read on how much the structural critique and the independence critique are actually in tension with each other?
Also worth understanding: one of the most contested aspects of NRC regulation is the linear no-threshold (LNT) model, the assumption that any radiation exposure, no matter how small, carries some cancer risk proportional to dose. Trump’s May 2025 executive order directly challenged this, directing the NRC to reconsider its reliance on LNT. Critics of LNT argue it’s scientifically outdated and that it drives enormous compliance costs for risks that may be negligible. Defenders argue that challenging it opens a door to weakening protections that public trust in nuclear power depends on. This is not a simple call, and watching the NRC navigate it over the next few years will tell us a lot.
The reform avalanche: ADVANCE Act, Part 53, and the executive orders
The pace of change at the NRC since 2024 is genuinely remarkable. It’s worth laying out what has actually happened, because the timeline is easy to lose track of. 🚀
The ADVANCE Act — Accelerating Deployment of Versatile, Advanced Nuclear for Clean Energy — was signed into law by President Biden in July 2024 with near-unanimous Senate support. Among other things, it changed the NRC’s mission statement for the first time in the agency’s history, explicitly directing it to regulate in a manner that does not “unnecessarily limit” nuclear energy development. That’s not a small thing. For decades the NRC had treated consideration of nuclear’s benefits as outside its remit, focusing exclusively on risk minimization. The ADVANCE Act said: weigh both sides.
By December 2025, the NRC had met 30 of its 36 planned ADVANCE Act deliverables. Then, in March 2026, it finalized Part 53 — a risk-informed, technology-inclusive regulatory framework applicable to any reactor type, not just light-water designs. This is the first new licensing framework since 1989 and the first update to reactor licensing standards since 1956. Key practical changes Part 53 brings:
Performance-based criteria replace prescriptive technology-specific requirements
Common design reviews allow multiple sites using identical designs to be reviewed together
Design certification terms extended from 15 to 40 years
Factory fuel loading options for modular designs that move significantly prefabricated
Flexible siting criteria that open industrial sites, data center campuses, and remote locations
Then came President Trump’s executive orders in May 2025, including one requiring a “wholesale revision” of the NRC’s structure, directing it to reduce staffing (in consultation with DOGE), establish an expedited pathway for designs already validated by the DOE or Department of Defense, and adopt science-based radiation limits. A separate executive order set a target of expanding American nuclear capacity from 100 GW today to 400 GW by 2050 — a four-fold increase in 25 years.
New NRC Chairman Ho Nieh, appearing before the House Energy and Commerce Subcommittee in April 2026, described the agency’s new posture in plain terms: “Enabling is really a mindset. It’s not a shortcut. It’s not a compromise. It’s just how we fulfill our safety authorities to benefit the American people.”
The independence question no one can cleanly answer
Here is where I think the story gets genuinely complicated, and where I’d caution against any simple narrative — whether the cheerful “NRC is finally fixed” version or the alarmed “safety is being gutted” version. 🔬
ProPublica’s April 2026 investigation documented some deeply unsettling specifics. The Trump administration fired a Democratic NRC commissioner — something that raised direct legal questions about the independence of the commission structure. NRC lawyers withdrew from proceedings before the Atomic Safety and Licensing Board, citing “limited resources” — the first such withdrawal in over 20 years, the judge noted. Career staffers told ProPublica that DOGE officials had unusual internal influence and that some employees were afraid to voice dissenting views. The budget for the NRC is proposed to be cut by nearly $80 million in fiscal year 2027.
At the same time, it’s fair to ask whether all of that staff time was well spent. The NRC’s own internal data shows that since 2016, Commission voting timelines have lengthened dramatically, procedural timeliness goals are met in only a small minority of cases, and the coordination failures at the Commission level are genuinely measurable. Some of what’s being cut may be unnecessary process. Some of it may be necessary protection. The honest answer is that it’s probably both, and distinguishing between them requires exactly the kind of independent technical expertise that’s now under strain.
The March 2026 policy integrity report on the NRC’s legal authority over SMRs concludes that collaborative reform — among developers, the NRC, and users — is preferable to letting courts or executive orders determine the regulatory future of American nuclear power. That seems right to me. The nuclear industry has the most to lose if the reform effort discredits the safety case for SMRs. History suggests that major nuclear accidents don’t just harm the plant involved — they set the whole industry back by decades. No one in this sector should be comfortable with a regulator whose independence is genuinely in question.
So here’s the question that matters most for anyone tracking the SMR industry in 2026: Can the NRC move fast enough to license the first wave of commercial SMRs on a timeline that keeps developers solvent and investors interested — while maintaining enough institutional rigor that a serious incident doesn’t hand the anti-nuclear movement its most powerful argument in a generation?