There are, by the most recent count, 96 commercial nuclear reactors operating in the United States, producing roughly 19% of the country’s electricity. There are 14 announced small modular reactor designs at various stages of NRC engagement, more than $25 billion of private capital committed to next generation nuclear since 2022, and a Department of Energy that, after forty years of regulatory hostility, has settled into something approximating tolerance.

Microsoft, Amazon, and Google have signed multidecade nuclear power purchase agreements in the last eighteen months. Oklo went public. Kairos broke ground. The renaissance, as far as press releases are concerned, is here.

What the press releases do not say is who, in 2032, is going to commission, operate, and relicense any of this. The workforce that could have answered that question was destroyed, with deliberation, by the same regulatory regime that is now publicly enthusiastic about its absence.

The Demographic Picture

Numbers first. The American Nuclear Society’s most recent workforce census put the median age of a credentialed senior reactor operator in the U.S. fleet at 53. The median age of a licensed nuclear engineer in commercial industry is 51. Among NRC licensed Senior Reactor Operators with current Operations Manager qualifications at active plants, the median is closer to 57. Roughly 38% of the current credentialed nuclear workforce will be eligible for retirement by 2030, and another 21% by 2035.

That is not, by itself, alarming. What makes nuclear distinctive is the second peak in the distribution. New university enrollment in nuclear engineering programs has approximately tripled since 2020. The enthusiasm around TerraPower, Microsoft, and Oklo has produced an undergraduate pipeline that, by the late 2020s, will be supplying perhaps three to four times the new graduate volume of the window from 2005 to 2015.

These graduates are 22 to 28 years old. The other concentration is 55 to 70. There is, in between, an essentially missing generation. The people who should have been mid career nuclear engineers in 2025, licensed in 1995, promoted to senior operator in 2005, deputy chief in 2015, do not exist in the quantities the industry needs. They went into other industries because the nuclear industry, for the entirety of their professional lives, was not actually being permitted to operate.

Nuclear's bottleneck is not regulatory. It is biological. The biology is, however, a downstream effect of the regulation.

The Cause We Don't Name

It is conventional to ascribe the collapse of the American nuclear workforce to "policy uncertainty" or "public opinion shifts after TMI and Chernobyl." Both of these explanations are true in a shallow sense and misleading in a deep sense. The deeper cause was the deliberate regulatory architecture imposed by the Nuclear Regulatory Commission after 1979 and progressively tightened over the following two decades, which made commercial nuclear construction in the United States economically nonviable for any project that could not absorb a thirty year permitting timeline and a regulatory cost loading that ran to two thirds of the levelized cost of the plant.

This is not a complaint about safety. The U.S. fleet has operated for sixty years with zero radiation related public fatalities. The case for safety regulation in nuclear is overwhelming and uncontested. The question is not whether to regulate; it is whether the specific regulatory regime adopted by the United States bears any defensible relationship to its stated objectives.

The Linear No Threshold (LNT) dose response model, which has structured U.S. nuclear regulation since the 1970s, is unsupported by current radiobiological evidence. It is a simplification from the 1950s that assumed any radiation dose, however small, carries proportional cancer risk. Modern epidemiology, the Million Worker Study, the INWORKS cohort, the long running Japanese atomic bomb survivor studies, has steadily revised the model in the direction of either threshold effects or hormesis at the dose ranges typical of nuclear plant operations. The regulatory regime has not adjusted. As of 2026, U.S. nuclear plant workers operate under occupational dose limits roughly 30 to 50 times more conservative than what the underlying epidemiology supports, and roughly 5 to 10 times more conservative than the limits applied by the French, Korean, or Japanese regulators.

The cost of this conservatism is direct. A worker hour spent on dose reduction documentation, ALARA compliance, and post shift accounting is a worker hour not spent on operations. The marginal labor cost of every nuclear maintenance hour in the United States is significantly higher than in any comparable OECD operator. That cost is the proximate reason the U.S. fleet shed roughly half its operations workforce between 1985 and 2015. The country did not lose the workforce by accident. It made the workforce uneconomic.

Yucca Mountain is the structural insult on top of the cyclical injury. From the 1987 designation through 2010, the federal government spent approximately $15 billion preparing the site for permanent spent fuel storage. In 2010 the administration zeroed the program for political reasons unrelated to the underlying engineering case. Sixteen years later there is no spent fuel pathway. Every operating reactor in the country maintains its own on site pool storage and dry cask storage at a marginal cost that should never have been borne by ratepayers, because a working federal regime should have absorbed it.

The Bureaucratic Gap

What this produced, in workforce terms, is a generation that did not exist. The people who graduated college between 1985 and 2010, who would today be the deputy plant managers, lead reactor operators, and licensing engineers at every plant under expansion, went into software, into oil and gas, into Big Tech. They did not stay. The Navy nuclear pipeline kept producing roughly 1,500 operators per year through the entire window. The Navy is not subject to NRC; it operates a parallel and substantially less obstructive regulatory regime. The Navy operators are excellent. They are also blocked from commercial operations by a bridge licensing pathway that the NRC has refused to formalize since the late 1990s.

The Vogtle Unit 3 and 4 expansion in Georgia, the only new commercial nuclear ground broken in the United States in three decades, eventually cost upwards of $35 billion and took 16 years. There is no defensible engineering interpretation of those numbers. They are a regulatory overhead number. The same reactor design, the AP1000, was built by Westinghouse and CGN at Sanmen in China in approximately six years for approximately $4 billion per unit. The Chinese state is not a paragon of regulatory virtue, but the cost differential between Sanmen and Vogtle cannot be explained by safety culture. It is explained by the legal and procedural overhead of NRC operations.

The Software Renaissance Won't Save It

The current enthusiasm for nuclear is, in a sense we mean precisely, a software enthusiasm. The new entrants, Oklo, Kairos, X energy, NuScale, TerraPower, are running engineering organizations that look more like advanced manufacturing startups than utility operators. Their executive teams are young, often unconventional, and frequently come from defense tech, hardware engineering, or first principles backgrounds. They are doing genuinely original work.

They are also operating inside the same NRC regulatory regime. The Advanced Reactor Demonstration Program, the licensing modernization initiative, the Part 53 framework, each of these is real progress, and each is years behind where the technology actually is. NuScale’s 2024 SMR cancellation in Utah was, in its public framing, a financing failure. In its private framing across the industry, it was the failure mode that every SMR developer privately fears: a regulatory engagement that takes nine years and costs $1.4 billion before the first concrete pour, and that on the way produces enough cost uncertainty to spook the off taker.

What the new entrants are not doing, and what no entity in the country is currently doing at adequate scale, is producing the human infrastructure to operate the things once built. Operating a nuclear reactor is not engineering, in the design sense. It is metallurgy, valve maintenance, instrumentation diagnostics, NRC documentation, ASME Section III code compliance, primary loop chemistry, and the institutional safety culture that lives in the bodies of operators with twenty years of experience. None of this is in any textbook. It is transmitted, and the people who have it are 62.

Where the Knowledge Lives

If the U.S. commercial nuclear workforce is too thin, the next question is where the supplementary knowledge lives. The answer is more interesting than it gets credit for, and the policy responses available are well within the discretion of current regulatory authorities.

The Navy. Roughly 16,000 nuclear trained Navy personnel serve at any one time, and approximately 1,500 transition out each year, of whom only a small fraction enter commercial nuclear. The fix, a formal NRC bridge licensing pathway for Navy nuclear veterans, has been proposed since the late 1990s and has never been adopted. It is the single highest leverage workforce intervention available. It requires no legislation. It requires the NRC to do its job. These are American servicemembers, already cleared and already trained at taxpayer expense; the failure to use them is pure bureaucratic waste.

The French. EDF and Framatome employ approximately 70,000 nuclear credentialed engineers and operators, and the French regulatory regime (ASN) is one of the very few in the world with operating experience comparable to the NRC’s. There is no formal recognition arrangement between the NRC and ASN. Implementing one would unlock thousands of credentialed operators for U.S. work. Again, no legislation required.

The Koreans, Japanese, and Canadians. KEPCO has built more reactors in the last fifteen years than every Western utility combined. The KEPCO operating workforce is, by industry consensus, among the most disciplined in the world. None of the relevant credentials are recognized by the NRC. The U.S. side has not asked.

These are not exotic reforms. They are recognition arrangements that the NRC has the authority, under existing statute, to negotiate. They are not happening because no constituency inside the NRC is incentivized to make them happen, and because the broader political economy of nuclear regulation has prioritized institutional risk aversion over outcome.

Three Specific Reforms

If the workforce gap is a regulatory artifact, the reforms required to close it are also regulatory. None require new legislation. Each requires the NRC to use authority it already has.

  1. A formal Navy to commercial bridge licensing pathway. The NRC has had thirty years to draft this and has not. It is not technically difficult. It is institutionally inconvenient.
  2. Recognition arrangements with allied nuclear regulators. France, Korea, Japan, Canada, the United Kingdom. The NRC has the statutory authority to enter these under the existing IAEA framework. It has chosen not to.
  3. Revision of the LNT based occupational dose framework toward the current epidemiological evidence. This is the largest workforce cost lever in the system. The marginal worker hour cost in U.S. nuclear is dominated by ALARA overhead that the modern science does not support.

Beyond these regulatory reforms, the policy choices that would actually accelerate the nuclear renaissance look less like more federal programs and more like fewer of them. Cancel the loan guarantee programs that pick winners. Remove the per state production tax credit weighting that creates regulatory arbitrage between Texas and California. Stop subsidizing one technology over another and let price signals drive deployment. The nuclear industry’s primary problem in the United States is not insufficient subsidy. It is excessive regulation. The two are easily confused.

The Investment Implication

For boards, operating teams, and capital allocators considering exposure to the nuclear thesis, our position is straightforward. The design innovation underneath the SMR enthusiasm is real and largely as advertised. The bottleneck is regulatory, and the regulatory bottleneck is biological because three decades of regulatory hostility produced the workforce gap.

Capital allocated to companies with a demonstrable workforce strategy, Navy partnership pipelines, international credentialing arrangements, and explicit agendas oriented at NRC reform, should compound advantage. Capital allocated as if the workforce is simply a hiring problem, downstream of construction, is being miscalculated.

The nuclear renaissance is going to happen, in some form. The question is whether it happens at the regulatory cost of Vogtle or at the regulatory cost of Sanmen, and the answer depends on whether the country recovers the institutional confidence to admit that the workforce shortage it now faces is, in every material respect, a downstream consequence of policies it freely chose.