Why It Takes So Long to Approve a Nuclear Reactor — And How That's Changing
The U.S. just rewrote its nuclear rulebook for the first time since 1989 — here's why it took so long, and what changes now.
Imagine pitching a business idea and being told you’ll get an answer in, oh, a decade or so. That’s been the rough reality for anyone trying to get a nuclear reactor approved in the United States. The process has historically taken years of pre-application work, years of formal review, public hearings, mandatory hearings, environmental assessments, and then more hearings. By the time a decision comes, the world has often moved on.
That slowness isn’t entirely accidental, and it isn’t entirely the NRC’s fault. But it has become genuinely unsustainable at a moment when the energy grid needs firm, zero-carbon power and AI data centers are consuming electricity at a pace that would’ve seemed absurd five years ago. The good news, if you’re willing to squint, is that something real is changing. Washington has rewritten the nuclear regulatory rulebook in a way it hasn’t since Ronald Reagan was in office. Whether those rewrites actually speed things up is a different question — and the answer is probably “yes, but not as fast as anyone wants.”
Why the old system took so long
The starting point is understanding just how the Nuclear Regulatory Commission built its licensing framework. The foundational rules, known as 10 CFR Part 50, date to 1956, when the Atomic Energy Commission was essentially writing the book on commercial nuclear power in real time. Those rules were designed around one type of reactor: the large light-water reactor (LWR), a pressurized or boiling-water design that became the American standard.
The problem is that most of the exciting reactor designs today — sodium fast reactors, molten salt reactors, high-temperature gas-cooled reactors, and the SMR variants that use them — are not light-water reactors. They don’t match the assumptions baked into Part 50. 🔬 When a developer submitted an application for a novel design, NRC staff essentially had to figure out, from scratch, how existing rules applied to something those rules were never meant to address. That’s a process that can eat years just in the preliminary stages.
Then there’s the sheer layering of the review itself. A typical new reactor application involves:
A pre-application engagement phase, often lasting 2–4 years, where the developer shares white papers and design documents with the NRC to surface issues before formal filing
An acceptance review, where the NRC decides if the application is complete enough to even start the clock
A safety review, producing a Safety Evaluation Report — which historically took 27 months or more
An environmental review under the National Environmental Policy Act, running in parallel but producing its own mountain of documentation
An Advisory Committee on Reactor Safeguards review, an independent layer on top of NRC staff’s work
A mandatory hearing, required by law even if nobody contests the application, plus additional contested hearings if public intervenors show up
The NRC’s generic milestone schedule for a new reactor application originally ran to roughly 39 months just for the formal review phase, on top of years of pre-application work. ⚡ And that’s the optimistic version. In practice, contested hearings have dragged approvals out far longer. Think about that: the Vogtle Units 3 and 4 combined license application was submitted in 2008, and those reactors didn’t reach commercial operation until 2023 and 2024 respectively.
Have you ever tried to read an NRC licensing application? They run to tens of thousands of pages. The AP1000 applications for Vogtle filled more than 100 volumes. That’s not bureaucratic waste — a lot of it reflects genuine technical complexity — but it does reflect a system that has never been designed with speed as a design criterion.
The architecture of delay: three structural problems
It would be easy to blame regulators, but the slowness has deeper structural causes. Three are worth understanding clearly.
First, the “first-of-a-kind” problem. Every novel reactor design enters the NRC process without precedent. There are no prior safety evaluations to reference, no operating experience to draw on, no established regulatory guidance tailored to the technology. 🧬 The NRC essentially has to build the review methodology and conduct the review simultaneously. According to the Nuclear Innovation Alliance.pdf), licensing timelines and costs have remained “uneven, often attributable to inconsistent quality in mundane but important practices like disciplined project management and clear internal and external communication.” Translation: even when the rules are clear, execution has been sloppy.
Second, regulatory mismatch. As Wikipedia’s entry on small modular reactors bluntly notes, NRC rules were built around reactors with electrical output above 700 MWe. SMRs often target outputs of 50 to 300 MWe. The staffing requirements, emergency planning zones, and safety analysis methodologies developed for behemoth plants don’t automatically scale down. That creates friction at every step. Some issues:
Multi-module SMR configurations (say, six reactors on one site) introduce new control room staffing questions the old rules never contemplated
Emergency planning zones written for large reactors may be excessive for smaller, inherently safer designs with lower decay heat
Factory fabrication of reactor modules introduces supply chain and vendor oversight questions Part 50 never addressed
Non-light-water coolants — liquid metal, molten salt, gas — require entirely new safety analysis frameworks
Third, the fee and workforce problem. The NRC charges applicants for its review time at a rate that, before recent reforms, ran to $318 per hour per staff-hour. A complex review requiring tens of thousands of staff-hours can cost an applicant tens of millions of dollars in NRC fees alone, before a single piece of steel is ordered. That’s a real deterrent. Small startups with novel designs often can’t sustain the financial bleeding of a multi-year review. This explains, partially, why Oklo’s first license application was rejected in early 2022 before the NRC even began a technical review — the application was incomplete, but the company also lacked the resources to sustain the depth of engagement the process demands.
What just changed — and it’s a lot
Here’s where things get interesting. 🚀 The last two years have produced more structural change to U.S. nuclear licensing than the previous three decades combined.
In July 2024, President Biden signed the ADVANCE Act — the Accelerating Deployment of Versatile, Advanced Nuclear for Clean Energy Act. It passed with broad bipartisan support, which in today’s Washington is itself remarkable. The ADVANCE Act did several concrete things:
Required the NRC to establish expedited procedures for qualifying new reactor applications
Cut NRC’s hourly fee rate for advanced reactor applicants to $148 per hour, a reduction of more than 50% from the standard rate
Authorized hiring incentives to address NRC’s chronic workforce shortages
Established prizes to cover licensing costs for early movers in novel reactor categories
Mandated that adjudicatory hearings be completed within specific timeframes
Then, in May 2025, President Trump signed Executive Order 14300, directing the NRC to establish fixed deadlines — actual deadlines, not guidelines — of 18 months for new reactor applications and 12 months for license renewals. The NRC revised its milestone schedules to reflect those targets, effective May 23, 2025.
Then came the real headline: on March 25, 2026, the NRC finalized Part 53 — a completely new regulatory framework, the first new reactor licensing rules since Part 52 in 1989, and the first major update to licensing standards since 1956. Part 53 is technology-inclusive, meaning it applies to any reactor type, not just light-water designs. It replaces prescriptive rules with performance-based criteria, allowing developers to demonstrate safety through probabilistic risk assessments rather than checking boxes designed for a different era. According to the American Action Forum, the new rule addresses the central problem: only two reactors approved under the 1989 framework ever reached commercial operation in nearly 40 years.
Does Part 53 actually deliver 18-month reviews? Jeremy Bowen, the NRC official leading the new Office of Advanced Reactors, expects reactor designs to receive approval in 18 months or less under the new framework. A 2023 NRC draft regulatory analysis estimated that, for just a single applicant, the cost savings could run between $53.6 million and $68.2 million. That’s not nothing.
The proof is already in the early data
It’s tempting to be skeptical — Washington has promised nuclear streamlining before. But there are genuine early signals that the culture is shifting. 💡
The NRC’s review of TerraPower’s Kemmerer Unit 1, a sodium fast reactor in Wyoming, ran for 18 months and finished nine months ahead of its original schedule and 11% under budget. The NRC also completed the safety evaluation for Kairos Power’s Hermes 2 test reactor four months ahead of schedule, partly by leveraging lessons from the Hermes 1 review to avoid redundant work. These aren’t just feel-good statistics — they represent a meaningful shift in how the agency approaches review efficiency.
That said, let’s be honest about what these wins don’t solve:
TerraPower still needs a separate operating license before it can actually run Kemmerer, meaning the project isn’t approved in any final sense yet
The overall Kemmerer project is running three years behind its original DOE target, with completion now targeted for 2030
Part 53 is optional — developers can still use the old frameworks — which may fragment the regulatory landscape rather than consolidate it
International regulatory fragmentation remains a serious problem; an SMR design licensed in the U.S. has to go through an entirely separate process in Canada, the UK, or anywhere else, undermining the economics of standardized factory production
The IAEA’s Nuclear Harmonization and Standardization Initiative is working on this international problem, and the analogy sometimes floated — that nuclear licensing should eventually resemble commercial aviation certification, where a design approved in one country gets recognized broadly — is appealing. But that’s a 10-to-20-year project, not a 2026 fix.
What “faster” actually means for SMR deployment
Here’s the uncomfortable math: even if the NRC hits its new 18-month target consistently, an SMR developer still needs 2–4 years of pre-application work before the clock even starts. Add 18 months of formal review, then construction (itself 3–5 years for a first-of-a-kind plant), and you’re still looking at a decade from serious commitment to electrons on the grid. 🌱
That’s not a counsel of despair. A decade is a lot better than what Vogtle demonstrated. And as the Nuclear Innovation Alliance’s 2024 reform report.pdf) argues, the real acceleration comes with nth-of-a-kind deployment — where a design is already certified and a new plant simply references that certification rather than relitigating the entire safety case. NuScale’s VOYGR design, the first SMR fully licensed in the U.S., provides a template: future NuScale applicants should face dramatically shorter reviews because the design itself is already on the books.
The question worth sitting with is this: given that the energy transition is happening now — grid operators are projecting demand surges from data centers, manufacturing reshoring, and EV adoption that won’t wait for a decade-long nuclear pipeline to mature — does even a reformed licensing process move fast enough? Or does it simply mean that the reactors powering 2035 are approved on time, while the reactors needed in 2030 still get stuck in the old system’s backlog?
What’s your read: is the NRC’s transformation real enough, and fast enough, to matter for this decade’s energy challenges — or are we still a generation away from nuclear actually filling the firm-power gap?