Why Does Nuclear Power Plant Construction Cost So Much?

The source of clean energy shouldn't be displaying "negative learning" over time
May 1st 2023

What does it cost to build nuclear power?

Nuclear plant construction is often characterized as exhibiting “negative learning.” That is, instead of getting better at building plants over time, we’re getting worse. Plants have gotten radically more expensive, even as technology has improved and we understand the underlying science better. 

Nuclear power currently makes up slightly less than 20% of the total electricity produced in the U.S., largely from plants built in the 1970s and 80s. People are often enthusiastic about nuclear power because of its potential to decarbonize electricity production, produce electricity extremely cheaply and reduce the risk of grid disruption from weather events. 

But U.S. nuclear power has been hampered by steady and dramatic increases in plant construction costs, frequently over the life of a single project. In the 1980s, several plants in Washington were canceled after estimated construction costs increased from $4.1 billion to over $24 billion, resulting in a $2 billion bond default from the utility provider. Two reactors being built in Georgia (the only current nuclear reactors under construction in the U.S.) are projected to cost twice their initial estimates, and two South Carolina reactors were canceled after costs rose from $9.8 billion to $25 billion. Why are nuclear construction costs so high, and why do they so frequently increase? Let’s take a look.

Nuclear power plants cost more and more

The story of nuclear power plants in the U.S. is one of steadily rising costs to build them. Commercial plants whose construction began in the late 1960s cost $1000/KWe or less (in 2010 dollars); plants started just 10 years later cost nine times that much. Today the Vogtle 3 and 4 reactors are likely to come in at around $8000/KWe in overnight costs ($6000/KWe in 2010 dollars), with an actual cost of nearly double that due to financing costs.

via Lovering 2016

The U.S. seems to do especially badly here, but most other countries have seen steadily rising construction costs. Here are French costs (which some experts have suggested is an underestimate):

And here are German and Japanese costs:

Most countries display a similar pattern of increasing costs into the 1980s, after which costs level off. The only country where the costs of construction seem to have steadily decreased is South Korea. South Korea’s outlier status has led some experts to speculate that the cost data (which the utility provides without an independent audit) may not be reliable.

Because nuclear plants take so long to build, we can observe cost increases over time at individual plants. A 1982 analysis of 75 U.S. nuclear plants found that cost estimates steadily increased as construction went on, with final construction costs two to four times as high as the initial estimated cost:

Most of nuclear energy’s costs come from construction

We can roughly break the costs of operating any power plant into three categories: fuel costs, operation and maintenance costs, and capital costs – the amortized cost from building the plant itself, including any financing costs.1 Different types of power plants have different cost breakdowns. For natural gas plants, up to 70% of their electricity cost comes from the cost of fuel. With nuclear plants, on the other hand, 60-80% of their electricity cost comes from constructing the plant itself. Decreasing the construction cost of plants would thus drive the cost of the electricity they provide down substantially.

From Dawson 2017

The cost fractions of different types of power plants (along with their technological capabilities) shape the way they’re used. Because electricity can’t be cheaply stored, at any given moment electricity produced and electricity consumed must balance. Since electricity consumption varies over time, power plants are brought on and offline as demand changes (this is called “dispatch”). The order in which plants are dispatched is generally a function of their variable costs of production (with lower-cost plants coming on first), as well as how easily they can ramp production up or down.2

Because capital costs make up the majority of the cost of nuclear power electricity, and those costs are largely fixed, nuclear plants tend to be operated continuously to supply “baseload” power; indeed, many plants in the U.S. are not even designed to ramp up and down easily.

Nearly all nuclear reactors globally are light water reactors: radioactive material in the reactor heats a supply of normal H2O or “light water,” which then transfers its heat to a second source of water, which then drives the turbine. A major risk for this type of reactor is a “loss of cooling accident.” If a coolant pipe bursts, or the supply of cooling water is otherwise disrupted and can’t cool the nuclear fuel, the fuel can heat up to the point where it melts down, damaging the reactor and potentially releasing radioactive material. Both the Fukushima and Chernobyl power plants experienced core meltdowns, and Three Mile Island experienced a partial core meltdown.

Even after the reactor is shut down, the radioactive materials continue to generate “decay heat” for an extended period of time. Thus even a reactor that has been heavily damaged needs to keep its cooling systems operational. Constructing cooling systems that can continue to operate in a damaged plant contributes heavily to nuclear construction costs. 

What’s contributing to high costs?

Inputs of plant construction costs

There are a variety of nuclear plant cost breakdowns available, but we’ll look at a breakdown done by the DOE in 1980 for a hypothetical 1100 MW plant, which should reflect the costs of U.S. plants, excluding financing, during the era when most plants were being built. 

(Greens are indirect costs, blues are direct costs)

Roughly one third of the costs are “indirect” costs of the building process: engineering services, construction management, administrative overhead, etc. For the direct costs (the costs of materials, equipment, and on-site labor), the reactor, the turbine equipment, and the plant structures each make up 15-20% of overall costs, with the balance made up by additional plant systems. Also note that the plant’s engineering design cost nearly as much as the reactor itself.

Rising labor costs are the bulk of increased construction costs

Most nuclear plant cost increase in the 1970s-80s can be attributed to increased labor costs. An estimate by United Engineers and Constructors found that from 1976-1988, labor costs for plant construction climbed 18.7% annually, while material costs escalated by only 7.7% annually (against an overall inflation rate of 5.5%) Of those labor costs, over half were due to expensive professionals: engineers, supervisors, quality control inspectors, and so on.

Other estimates align with this. A 1980 estimate produced by Oak Ridge suggests that material volume increases between the early 1970s and 1980 generally ranged from 25-50%, not nearly enough to account for the total cost increases seen:

And a recent paper by Eash-Gates et al examined cost increases for a sample of nuclear power plants built between 1976-1988. It found that 72% of the cost increase was due to indirect costs, indicating a large increase in expensive professionals such as engineers and managers:

Similar issues seem to be taking place today. An OECD NEA study that looked at nuclear plant cost estimates made from 2010-2020 found that increases in “indirect cost are the main driver of [nuclear plant] cost overruns,” with labor making up 80% of indirect costs.

via OECD

Why did labor costs increase? According to most observers, increasing regulation made plants increasingly burdensome to build. During the late 60s and early 70s, regulatory requirements steadily increased:

Regulatory guides issued by the NRC, from Szalay 1978

As did the thoroughness of review by the Nuclear Regulatory Commission (NRC), which is responsible for issuing plant operating licenses:

via Szalay 1978

 A 1980 study found that increased regulation between the late 1960s and mid 1970s was responsible for a 176% increase in plant cost, and increased labor requirements by 137%:

And the Eash-Gates study found that at least 30% of the cost increase between 1976-1988 can be attributed to increased regulation. For a vivid overview of the impact of increased regulation, see Charles Komonoff’s 1981 “Power Plant Cost Escalation”:

“One key indicator of regulatory standards, the number of Atomic Energy Commission (AEC) and Nuclear Regulatory Commission (NRC) “regulatory guides” stipulating acceptable design and construction practices for reactor systems and equipment, grew almost seven-fold, from 21 in 1971 to 143 in 1978. Professional engineering societies developed new nuclear standards at an even faster rate (often in anticipation of AEC and NRC). These led to more stringent (and costly) manufacturing, testing, and performance criteria for structural materials such as concrete and steel, and for basic components such as valves, pumps, and cables. 

Requirements such as these had a profound effect on nuclear plants during the 1970s. Major structures were strengthened and pipe restraints added to absorb seismic shocks and other postulated “loads” identified in accident analyses. Barriers were installed and distances increased to prevent fires, flooding, and other “common-mode” accidents from incapacitating both primary and back-up groups of vital equipment. Similar measures were taken to shield equipment from high-speed missile fragments that might be loosed from rotating machinery or from the pressure and fluid effects of possible pipe ruptures. Instrumentation, control, and power systems were expanded to monitor more plant factors under a broadened range of operating situations and to improve the reliability of safety systems. Components deemed important to safety were “qualified” to perform under more demanding conditions, requiring more rigorous fabrication, testing, and documentation of their manufacturing history. 

Over the course of the 1970s, these changes approximately doubled the amounts of materials, equipment, and labor and tripled the design engineering effort required per unit of nuclear capacity, according to the Atomic Industrial Forum.”

The 1979 accident at the Three Mile Island nuclear plant in Pennsylvania accelerated this trend. Required safety changes added an estimated 10% to the labor costs and 15% to the material costs of a new plant. Eash-Gates found that the rate of material deployment, or how fast construction materials are installed on site, showed “a precipitous drop between 1979 and 1980 following the Three Mile Island accident.”

Material deployment rates on nuclear plants over time

Some regulatory increase wasn’t necessarily unreasonable. Early safety requirements for nuclear plants often overlooked critical risks. For instance, prior to the proposed power plant at Bodega Bay near the San Andreas fault, seismic activity had not been considered in the design of nuclear plants. Subsequent analysis revealed that the potential for severe seismic events was much more widespread than had previously been thought. Similarly, tornado design requirements weren’t created until an application to construct a plant in a high tornado area revealed that tornado risk was much more widespread than had been assumed. When accident risk was considered, it was often analyzed incorrectly. A reactor meltdown was thought to be an astonishingly unlikely accident, yet Three Mile Island experienced one after relatively few reactor-years of operation.

Regulations constantly change

In response to learning more about how nuclear reactors could fail, the NRC’s regulatory stance became a deterministic, defense-in-depth approach – the NRC imagined specific failure modes, and specific ways of preventing them, and then tried to layer several redundant systems atop each other to compensate for uncertainty. Whenever something new was learned about potential failure modes, the regulations were changed.

These changes applied not only to future plants, but often to plants under construction. In some cases existing work had to be removed, requiring intervention and oversight from design engineers, managers, field inspectors, and other expensive personnel. This became another major source of increased costs. Here’s Komonoff again: 

“…because many changes were mandated during construction as new information relevant to safety emerged- much construction lacked a fixed scope and had to be let under cost-plus contracts that undercut efforts to economize. Completed work was sometimes modified or removed, often with a “ripple effect” on related systems. Construction sequences were frequently altered and schedules for equipment delivery were upset, contributing to poor labor productivity and hampering management efforts to improve construction efficiency.”

A universal tenet of large construction projects is that one should avoid changing the design during construction. Changes while a project is in-progress may require existing work to be removed, or new work to be done in difficult conditions. It often requires significant coordination effort just to figure out what work has been already done (“Have you poured these foundations yet? Are the columns in yet?”). If a pipe needs to run through a beam, it’s easy to design the beam ahead of time to accommodate it. But if the beam has already been fabricated, you might have to field-cut a hole, or add reinforcing. Or maybe the beam can’t accommodate the hole at all, and you need to redesign the entire piping system (which will of course impact other in-progress work). While this expensive redesign is happening, everyone else might need to stop their work. 

This sort of disruption is especially costly on a construction project the size of a nuclear plant. A nuclear plant can employ up to 5,000 construction workers at a time. As Komonoff notes, “Reactors in the 1970s were built increasingly in an ‘environment of constant change’ that precluded control or even estimation of costs, and which magnified the direct cost impacts of new regulations and design changes.” Constant changes made it nearly impossible to coordinate site work effectively. A 1980 study of nuclear plant craft workers found that 11 hours per week were lost due to lack of material and tool availability, 8 hours a week were lost in coordination with other work crews or work area overcrowding, and 5.75 hours per week were lost redoing work. All together, nearly 75% of working hours were lost or unproductively used. This trend accelerated following Three Mile Island, as updated safety requirements after the accident once again had to be implemented on in-progress plants at great cost.

The Eash-Gates study finds that costs have steadily increased even for “standard” reactor designs. In 2009, for instance, Westinghouse was forced to change the containment building for its AP1000 reactor to withstand aircraft strikes (a post-9/11 ruling by the NRC), seven years after it had applied for approval of the design. The subsequent change, which had to be implemented on the in-progress Vogtle and VC Summer plants, has been blamed for delays and cost increases on the two plants. A 1978 presentation from a member of the Atomic Industrial Forum argued that ”achieving stable licensing requirements is the clear target for any effort to obtain shorter and more predictable project durations.” 

Constant regulatory change also imposes a coordination cost on builders and regulators, as they gradually work their way to a mutual understanding. A study on nuclear power craft productivity describes the issue:

“Varying interpretations of the plans, specifications and building code stipulations among quality control inspectors was another frequent occurrence, according to the craftsmen. Each of these predicaments was said to cause continual postponements and removal and reinstallation of work that was deemed to be non-conforming.”

And similar issues seem to be behind cost overruns at the Flamanville plant in France. New requirements were imposed while large components were being manufactured for the Flamanville EPR, requiring industry and certification bodies to reinterpret and agree upon the evolving requirements and lengthening the construction time by years.

Partially because of continual regulatory change, the time required to build a nuclear plant in the U.S. has continuously increased. The minimum time required to build a plant increased from 4 years in the late 60s to 8 years in the mid-1970s, with 75% of reactors taking 10-15 years to build. Vogtle Units 2 and 3 will have taken more than 14 years to complete, assuming work wraps up this year. Some of this increase was the result of the Calvert Cliffs court case, which mandated that an environmental impact review must be performed for every plant built. This extended construction duration adds even more cost. Long project durations increase financing and labor costs, as well as the probability that new regulations, objections, or other blockers will cause further obstacles.

To summarize so far: In the U.S., labor costs increased dramatically, especially labor from expensive professionals. This labor cost increase was at least partly due to frequently changing regulations during the period, which caused extensive design changes, delays, rework, and general coordination issues on in-progress plants.

Quality Assurance/Quality Control requirements are incredibly onerous

In addition to generating substantial increases in labor costs, regulations also influence the direct costs of nuclear plant construction via QA/QC requirements. Plant components require extensive testing and verification to ensure they’ll continue to function even after extreme accidents. This often takes the form of carefully recording what happens to every component at each step of the manufacturing and construction process, to ensure the correct part with precise performance characteristics is put in the right place.

This sort of documentation can be extremely burdensome to create. For example, here’s a description of QA requirements during the construction of the Diablo Canyon nuclear plant, via Komonoff again: 

“Simple field changes to avoid physical interference between components (which would be made in a conventional plant in the normal course of work) had to be documented as an interference, referred to the engineer for evaluation, prepared on a drawing, approved, and then released to the field before the change could be made. Furthermore, the conflict had to be tagged, identified and records maintained during the change process. These change processes took time (days or weeks) and there were thousands of them. In the interim the construction crew must move off of this piece of work, set up on another and then move back and set up on the original piece of work again when the nonconformance was resolved… Every foot of nuclear safety-related wire purchase is accounted for and its exact location in the plant is recorded. For each circuit we can tell you what kind of wire was used, the names of the installing crew, the reel from which it came, the manufacturing test, and production history.”

Similar documentation requirements apply to the manufacture of nuclear-grade components. A former engineer describes the process:

“Many moons ago I did [design and manufacturing] for a company that made both (section VIII and section III  [nuclear] vessels) and my memory is that it was essentially the same design work with much more documentation and paperwork required for the ‘N’ stamp vessel. When the paper weighed about what the vessel did, it was ready to ship.”

Dawson 2017 estimates that quality control requirements make up 23% of the cost of concrete, and 41% of the cost for structural steel on nuclear plants:

An analysis by EPRI found that nuclear-grade components were in some cases 50x more expensive than off-the-shelf industrial grade ones. 

Nuclear-grade components don’t necessarily have higher performance requirements than conventional components. Reinforcing steel in nuclear-grade concrete, for instance, is the same material used in conventional concrete. Instead, the additional cost often comes from the additional documentation and testing required. Documentation requirements also increase costs indirectly, by reducing market competition among manufacturers. Because these requirements are difficult for manufacturers to implement, many simply don’t bother to manufacture nuclear-grade components. Combined with the fact that the US spent a long period of time not building new nuclear plants, this limits the pool of potential nuclear component suppliers, making it harder to obtain components and further increasing their price. Some experts think these QA/QC requirements and their downstream market effects are the prime reason for high nuclear construction costs:

“…the main factor leading to high plant construction costs is not the design of the reactors, or various safety features that they employ, but the uniquely strict QA requirements that apply (only) for the fabrication of safety-related nuclear plant components (i.e., ‘nuclear-grade’ components). Conversely, I believe that in terms of safety, fundamental reactor design, employed safety features, intelligent operation/training, and maintenance are much more significant (effective) than the application of extremely stringent fabrication quality control requirements.”

The difficulty of QA/QC create a drain on technical ability

Because the requirements for constructing nuclear plants are so strict, in practice it’s often very difficult for builders and manufacturers to meet them. Consider concrete, a material required in significant amounts for plant foundations, as well as the containment building. Because it performs a shielding function, nuclear plant concrete must meet the stringent safety-related QA/QC and documentation requirements. At the VC Summer plant in South Carolina, a concrete work package took three volumes of documents. According to the Bechtel Project Assessment Report

One volume [has] safety bulletins, quality control sign-off sheets, and general information associated with the work, one has drawings and specifications, and one has design changes. In some packages, the design change volume is twice as thick as the drawing volume.”

Meeting these requirements for a site-produced material is difficult. Nuclear concrete typically has multiple closely-spaced reinforcing bars that can be difficult to arrange properly (the Royal Academy of Engineer’s 30-page Guide to Nuclear Concrete mentions “congestion” 13 times). Concrete placement issues have plagued every recent nuclear project and are frequently the source of delays and cost overruns. Examples abound: a 6 month delay from incorrectly placed rebar on Vogtle 3 and 4 in Georgia, a 4 month delay on the VC Summer plants for similar reasons, and a 9-month delay from poor concrete composition at the Olkiluoto 3 in Finland.

The difficulty of meeting requirements, combined with the lack of construction expertise due to long periods without constructing new plants, means that any new construction inevitably struggles as the builders learn how to meet the high level of stringency required. Delays at Vogtle Units 3 and 4 were partially due to a contractor unprepared for the difficulty of nuclear construction. Similar issues seem to be responsible for delays and cost overruns on Flamanville in France and Olkiluoto in Finland.

How can we bring costs down?

What hasn’t worked yet?

Most attempts to improve the U.S. nuclear plant construction process don’t seem to have worked.

One major attempt was a change to the plant licensing process, which originally involved two steps. Applicants would first apply for a construction license by providing a basic safety analysis, which would allow them to start construction on the plant. The construction license didn’t require a fully specified plant design. Once the plant was complete, the operator would then apply for an operating license, allowing the plant to start producing power. Building a nuclear plant without knowing whether the design was acceptable was obviously a source of difficulty: it was this licensing structure, for instance, that was partly responsible for forcing in-progress plants to meet constantly changing regulatory requirements.

In the 1990s this was replaced with a 1-step licensing process, where applicants would provide a completely specified design as part of the application, and receive a combined construction and operating license (COLA). However, so far this change seems to have backfired. Because the design was already approved, any deviation is now required to go through several additional levels of approval, making it even harder to make on-site changes.

For instance, the VC Summer Units 2 and 3 and the Vogtle Units 3 and 4 were all permitted under the 1-step licensing process. These reactors all used the then-new AP1000 reactor from Westinghouse, which was designed to be much simpler than previous reactors, having “60% fewer valves, 75% less piping, 80% less control cable, 35% fewer pumps and 50% less seismic building volume than usual reactor design.” The AP1000 also has an emergency cooling system that works passively via gravity, and is thus (theoretically) less susceptible to loss of cooling accidents, since it doesn’t require power to operate. The reactor was also designed to be prefabricated and installed on-site in large modules, reducing the requirements for site labor.

However, the initial design of the AP1000 had significant constructability issues.3 The reduced footprint seems to have forced systems and components much closer together, making them difficult or impossible to install and requiring frequent design changes. There also were issues with the prefabricated modules, which were often behind schedule, out of spec, and requiring significant rework – one downside of prefabrication is that problems that occur on-site are more difficult to fix. The added hurdles for making on-site changes from the 1-step licensing process exacerbated all these issues.

Another common strategy for reducing plant construction cost is to build many identical copies of the same plant. It’s typically been assumed that first-of-a-kind (FOAK) plants will be more expensive, and that re-using the same design on future projects (nth of a kind, or NOAK, plants) will cause learning curve effects to reduce costs. But the Eash-Gates study found that this hasn’t occurred in the U.S., likely due in part to the frequently changing regulations. It doesn’t matter how standardized your design is if you end up needing to change it on every project to meet new requirements.

What might work?

It’s possible to build nuclear reactors without enormous costs. Assuming the data is legitimate, both China and Korea have managed to build several recent reactors in under six years without major cost overruns. The UAE’s recent Barakah reactor, which will be completed in 2023 by the Korea Electric Power Corporation, came in 25% over budget, compared to the Vogtle Units 3 and 4 now being completed in the U.S., which were more than 100% over budget. And while France’s recent nuclear plant construction performance is less than impressive, historically they achieved much better results. Between 1970-2000, France built 58 nuclear reactors, at significantly lower costs than U.S. reactors. French plants were also built an average of 2.6 years faster than U.S. plants.

French vs U.S. nuclear plant costs, via Grubler 2010

The OECD NEA report “Unlocking Reductions in the Construction Cost of Nuclear Power Plants,” which analyzed construction of dozens of plants around the world, breaks strategies for nuclear cost reduction into four categories: design and supply chain maturity, regulation stability and predictability, effective project management, and policy frameworks.

Design and supply chain maturity

The scale and lead time of nuclear plant construction often results in builders being pushed to start construction before the design is complete. However, proceeding with an incomplete design inevitably results in design changes during the process of construction, causing delays and expensive on-site rework. Lower-cost plants have greater percentages of their design completed at the start of construction than higher-cost plants.

To minimize the likelihood of cost overrun, plants should be built using mature designs that don’t need to be changed during the construction process. In construction of the French reactor fleet, for instance, the CEO of the French utility company EDF noted that during plant construction “Whenever an engineer had an interesting or even genius [improvement] idea either in-house or at Framatome, we said: OK, put it on file, this will be for the next series, but right now, we change nothing.”

By building multiple reactors using an unchanging design, the benefits of learning-by-doing can be unlocked. In the French reactor fleet, though costs increased whenever a new reactor type was introduced, later plants using a given reactor design tended to be cheaper than earlier plants. Similarly, the 4th unit built at the UAE’s Barakah plant was 50% cheaper than the 1st unit.

Long periods spent without construction of new plants also mean the nuclear supply chain withers, and workers’ and companies’ experience building nuclear plants is lost. Without companies with experience in building nuclear plants, new plant construction will inevitably be slower and costlier as competencies are re-acquired. Building nuclear plants cost effectively requires developing and maintaining an experienced nuclear workforce.

Regulation stability and predictability.

Reaping the benefits of design maturity and repetitive construction, however, requires regulatory stability. A stable design is only possible if a plant can be permitted and built without needing to be changed to conform to updated regulatory requirements. Changing the design of a plant during its construction in response to regulatory changes inevitably results in increased costs and project delays. And because the updated design isn’t in the original project scope, the change designs often require the use of “cost-plus” contracts, which reduce incentives for contractors to complete the work under budget.

Regulatory changes that inevitably do occur must be made predictable: it must be clear when they’ll be introduced, to what projects they’ll apply to when they are introduced, and exactly how the regulations will be translated to technical requirements. Plants under construction should be grandfathered in under earlier regulations.

Effective project management

A large fraction of the cost of a new nuclear plant is the cost of the labor to build it and the financing costs, both of which are strongly impacted by project delays and rework, making effective project management of nuclear plant construction essential. As with the nuclear supply chain, the lack of project management experience for nuclear plant construction (one of the most complex and expensive capital projects in existence) hinders delivering new plants on time and on budget. Nuclear plant project management skills must be maintained alongside the physical aspects of the supply chain. Project management is also aided by a mature plant design and a stable regulatory environment, which make construction issues less likely to occur and the schedule planning more feasible.

Project management can also be helped by properly structuring project contracts to incentivize cost-savings, while recognizing the risk inherent in nuclear plant construction. Fixed or firm-cost contracts, which place all the risk on the contractor, “come at the price of a significant premium due to caution on the part of EPC contractors,” according to the OECD NEA report. Risks that do arise tend to result in adversarial conditions between stakeholders rather than collaboration, as each party attempts to shift blame onto the other. This can be addressed with the use of contracts that have incentive payments for on-time, on-cost construction.

Policy frameworks

The above goals are all heavily influenced by government policy choices. A government enthusiastic about nuclear power and which commits to constructing a series of plants can enable cost savings from repeatable plant construction, reduce contractor risk, and help ensure that the nuclear supply chain remains robust. A government that is skeptical of nuclear power, and creates a morass of constantly shifting regulations and uncertainty, on the other hand, makes it impossible for nuclear power to be cost competitive.

Managing expectations: public support for nuclear

U.S. nuclear regulation stringency is sometimes blamed on the policy of “ALARA” adopted by the NRC: that radiation exposure to workers and the public should be “As Low As Reasonably Achievable.” Critics point out that strict interpretation of this philosophy results in ever-stricter regulations that prevent nuclear from ever being cost-competitive. ALARA, in turn, is based on a “linear no threshold” model of radiation safety that many now believe is incorrect.

In practice, the AEC/NRC do seem to have had a deliberate policy of creating increasingly strict regulations to minimize potential radiation exposure. But it’s easy to over-index on ALARA as a specific driver of high U.S. nuclear construction costs. Every country in the world has adopted the ALARA standard, as has the U.S. Navy, so ALARA can’t be the sole reason for high nuclear plant construction costs. And blaming ALARA suggests an overly simple causal chain of regulatory response. In particular, it omits the role of public concern and controversy in influencing regulations and decisions around nuclear energy, which historically has been a major factor: the word “controversy” appears 26 times in the NRC’s 116-page “Short History of Nuclear Regulation.”

For instance, negative public response was a major factor in the NRC abandoning attempts to reduce regulatory requirements for materials with low levels of radiation: 

“The agency proposed that if radioactive materials did not expose individuals to more than 1 millirem per year or a population group to more than 1,000 person rem per year, they could be [exempt from regulatory controls]. The NRC explained that the BRC policy would enable it to devote more time and resources to major regulatory issues and thereby better protect public health and safety.” 

The NRC’s announcement of its intentions on the BRC policy was greeted with a firestorm of protest from the public, Congress, the news media, and antinuclear activists. Critics suggested that BRC policy would allow the nuclear industry to discard dangerously radioactive wastes in public trash dumps. In public meetings held by the NRC, citizens called repeatedly for the resignation of the Commissioners or their indictment on criminal charges.” 

In response to the BRC policy, five states banned the disposal of nuclear waste with low levels of radioactivity in their landfills, and dozens of environmental groups filed lawsuits against the NRC. In response, the NRC declared a moratorium on the Below Regulatory Concern policy.

We see similar public safety concerns driving nuclear power policy in other countries. Public opinion turned against nuclear power in Germany following the 1986 Chernobyl accident, and following the 2011 Fukushima disaster most polled Germans supported phasing out nuclear power. Italy shut down all its nuclear reactors after Chernobyl, and a referendum to reintroduce them in 2011 (shortly after the Fukushima disaster) was voted down with 94% of the vote. While an improved regulatory framework for nuclear power is possible, it would require the public to accept significantly relaxed safety requirements.

Managing expectations: the naval example

The OECD recommendations listed above give a recipe for reducing the construction cost of nuclear power plants in the U.S.. However, we should be realistic about what these interventions would be able to achieve. A relevant comparison here is the U.S. Navy. The Navy has been building nuclear reactors for 70 years as part of its Naval Reactors program, and has built over 200 nuclear-powered ships and 500 reactor cores: that’s more experience building nuclear reactors than anyone in the world.

In many ways, the design philosophies of naval and civilian reactors are similar. Like civilian reactors, naval reactors are built with an overwhelming focus on safety. The Navy follows the ALARA philosophy, and goes to great lengths to minimize exposure to radiation. No military or civilian personnel in the Naval Reactors program have ever received more than 10% of the annual occupational exposure limit, and the average annual exposure of reactor personnel is 0.112 rem per year, compared to the 0.620 rem per year average of the U.S. population as a whole. Naval reactors have operated for over 5000 reactor years and over 130 million miles of travel without a major accident. Like the NRC, naval reactors also use a “defense in depth” approach to safety, with multiple overlapping, redundant safety systems to prevent single points of failure. The Navy also ensures performance requirements are met via an extremely thorough system of quality control and assurance.4 

Unlike civilian reactors, the Navy has often (though not always) managed to control costs and schedule on nuclear ship construction. For instance, the Virginia class submarine, which went into production following the fall of the Soviet Union, was deliberately designed to be inexpensive to produce, and to learn lessons from its more expensive predecessor, the Seawolf program. The program targeted a submarine inexpensive enough that the Navy could afford to build two per year, which would help maintain its industrial base. As part of this effort, the Virginia class deliberately eschewed untested technology as much as possible, and avoided trying to push the boundaries of submarine performance. Technology and systems were deliberately chosen on a cost-benefit approach. For the newly developed systems that it did include, the Navy used a “try before buy” acquisition strategy, where new equipment was tested first on land or other ships prior to construction. The Virginia program also emphasized upfront design work and attempting to resolve problems before they occurred during construction.

The first Virginia class cost ~$2.8 billion to build (in 2004 dollars), 50% more than originally budgeted for. Focus was thus placed on improving the construction process of subsequent subs. The builders managed to cut over 100,000 labor hours out of the construction process by simplifying design and increasing automation. The cost was eventually brought down to less than $2 billion in 2004 dollars. Though there have been some hiccups, construction times for Virginia subs have steadily decreased.

The strategies for successful cost and schedule management on nuclear submarine construction largely overlap with the OECD recommendations for improving civilian plant construction. Repeatability is emphasized: naval reactors show a level of design reuse that is often desired but seldom achieved in civilian plants. For instance, the S5W reactor was used on 98 different submarines of 8 different classes between 1959 and 1979. And though the Virginia class has undergone several design evolutions since its inception, all versions use the S9G reactor. Learning-by-doing/nth of a kind effects are achieved much more successfully in nuclear ship construction than in U.S. civilian reactor construction.

Similarly, it’s recognized that nuclear submarines (along with the reactors that power them) are incredibly complex artifacts, that only a few firms and people possess the required skills to build them, and those capabilities may deteriorate if they aren’t deliberately maintained. Emphasis is thus placed on maintaining the industrial base that can construct them. For instance, the second and third Seawolf subs were built largely to maintain continuity in the industrial base until the next submarine program began. And the Virginia submarine program was structured with both Electric Boat and Newport News building portions of each sub partly to ensure multiple shipyards maintained the ability to build nuclear submarines.

The damaging effects of changing design requirements and starting construction with an incomplete design are also recognized in nuclear sub construction. When construction on the first Virginia class sub began, design was 50% complete (compared to just 10% on the first Seawolf class). Design requirements were frozen early on to minimize changes required during construction.

However, this hasn’t made nuclear power an inexpensive submarine propulsion system. A nuclear-powered naval vessel is not, and has never been, less expensive to produce than a conventionally powered ship. A 1961 naval study estimated that a nuclear-powered surface ship would cost approximately 1.5 times what an oil fired powered ship would cost.5 Of that extra cost, approximately 1/3rd was from the nuclear fuel, with the rest the cost of the reactor and additional systems.

In 1982, it was estimated that a French nuclear submarine cost approximately 1.7 times what a conventional diesel electric submarine cost, and also had higher maintenance and support costs. More recently, a CRS report estimated that making a naval surface ship nuclear-powered would add $600-800 million to the cost of the ship in 2007 dollars ($845-$1,127 million in 2022 dollars).

This suggests that a nuclear power plant is many times more expensive than a conventional (oil fired, gas turbine, diesel, etc.) power plant on a ship. Beyond the costs of the ships themselves and the crew to operate them, fielding a nuclear force requires the Navy to maintain an organization of 5000+ people in the Naval Reactors organization and its research arms. The argument for nuclear ships in the Navy has never been “nuclear is a cheaper source of power” (though it theoretically could be if oil prices rose enough). Instead, proponents have argued the benefits of nuclear propulsion, namely unlimited operation at high speeds without worrying about refueling, are worth the additional costs.

The nuclear Navy follows almost all the recommendations of the OECD report, and paints a picture of what an improved nuclear industry might look like. We can build reactors for cheaper than we currently do. We can unlock the benefits of learning by doing, and complete nuclear plants on-time and on-budget. But a nuclear reactor is likely to remain an especially complex and expensive capital project, and at our current current technology and safety requirements is unlikely to result in significantly reduced electricity costs.


To sum up, since the early 1970s, the cost of constructing nuclear power plants in the U.S. has been steadily rising. This can be traced to a constantly shifting regulatory environment, which has continuously changed plant design requirements, and added more and more safety features, which often were required to be implemented on plants under construction. The regulatory environment is partially a reflection of the fact that nuclear power and the risks of radiation had become increasingly controversial, and that early understanding of the likelihood of a nuclear plant accident was often inadequate.

Constantly changing regulations continuously added costs in the form of more safety features, project delays and on-site rework. And they have prevented potential learning-by-doing gains from producing repeatable, standardized plant designs at scale. Plants were constantly late and over-budget, and became an increasingly unattractive financial investment. Starting in the late 1970s plants began to be canceled in large numbers, which accelerated following the accident at Three Mile Island. No new nuclear plant has been scheduled for construction in the U.S. since 1978. The nuclear supply chain has withered, making new plants even harder to construct. Recent attempts at nuclear plant construction have at best ended with massive budget overruns (in the case of Vogtle Units 3 and 4). At worst, they’ve ended in failure after billions were spent (in the case of VC Summer Units 2 and 3).

However, it’s not impossible to deliver nuclear plants in reasonable amounts of time for a reasonable budget. We have a playbook for improving this process. By using mature plant designs that can be built repeatedly, learning-by-doing gains can be achieved, making each plant built cheaper than the last. By developing and maintaining a robust nuclear supply chain with the necessary expertise and experience, we can ensure we don’t lose the ability to deliver plants in the future. By stabilizing regulations, making them clear, and making changes to them predictable, we can prevent cost overruns associated with expensive and time-consuming on-site rework.

But we should be realistic about what this playbook might achieve. Public concern about nuclear accidents likely makes any significant reduction in plant safety requirements untenable. Experience with the Navy’s nuclear program suggests that even by following the above playbook, building a nuclear plant to the level of safety required is a fundamentally expensive undertaking. Truly moving the needle on nuclear power might require a ground-up rethinking of how we build plants, towards things like small modular reactors or nuclear plants built in shipyards in large numbers and floated into position along the coast.

  1. Comparison between different types of power plants is often done using overnight costs, or the cost to build them if they were built overnight and did not have to pay interest charges. Because nuclear plants reliably take longer (sometimes much longer) to build than other types of power plants, this biases these types of comparisons in favor of nuclear plants. For nuclear plants with severe construction delays (such as the Vogtle 3 and 4 plants in Georgia), financing costs can be a significant fraction of the cost.

  2. This is another complicating factor in comparing different types of power plants. Comparison is often done by comparing the “levelized cost of electricity” (LCOE) of a plant, the net present cost of the electricity a plant will produce over its lifetime. But the value of the electricity a plant produces is different at different times. For instance, the price of electricity at night is often lower. By contrast, natural gas peaker plants will be used when demand is unexpectedly high, and the price of electricity is much higher. Intermittent sources such as wind and solar will sometimes produce more electricity than is expected or needed, which can push down the price, in some cases enough that the price actually goes negative. Nuclear, wind, and (sometimes) solar thus might often be selling less valuable electricity than other types of power plants.

  3. A frequent issue in nuclear plants, due to the amount of piping, wiring, and other services they require.

  4. See "Why are nuclear power construction costs so high? Part III - the nuclear navy" for a full accounting of the differences between civilian and naval reactors.

  5. By which time the U.S. had built close to 20 nuclear-powered subs and surface ships.