Most U.S. nuclear power plants cost more to run than they earn. Globally, the World Nuclear Industry Status Report 2019 documents the nuclear enterprise’s slow-motion commercial collapse—dying of an incurable attack of market forces. Yet in America, strong views are held across the political spectrum on whether nuclear power is essential or merely helpful in protecting the Earth’s climate—and both those views are wrong. In fact, building new reactors, or operating most existing ones, makes climate change worse compared with spending the same money on more-climate-effective ways to deliver the same energy services. Those who state as fact that rejecting (more precisely, declining to bail out) nuclear energy would make carbon reduction much harder are in good company, but are mistaken.
If you haven’t heard this view before, it’s not because it wasn’t published in reputable venues over several decades, but rather because the nuclear industry, which holds the microphone, is eager that you not hear it. Many otherwise sensible analysts and journalists have not properly reported this issue. Few political leaders understand it either. But by the end of this article, I hope you will. For the details and documentation behind this summary, please see pp. 228–256 of the World Nuclear Industry Status Report 2019. A supporting paper provides simple worked examples of how to compare the “climate-effectiveness” of different ways to decarbonize the electricity system.
Nuclear power’s potential role in the global climate challenge
If the nuclear one-tenth of global electricity generation displaced an average mix of fossil-fueled generation and nothing else, it would offset 4% of fossil-fuel CO2 emissions. So in an era of urgent climate concern, should nuclear power continue, shrink, or expand? Should its successive previous rationales—replacing insecure oil, then replacing dirty coal, then fighting global poverty—now add a fourth one, protecting climate?
Six months ago, a report by the International Energy Agency claimed that not sustaining and even expanding nuclear power would make climate solutions “drastically harder and more costly.” To check that claim, we must compare nuclear power with other potential climate solutions. What criteria should we use? Here I’ll use only two—cost and speed—because if nuclear power has no business case or takes too long, we need not address its other merits or drawbacks.
How should we compare different ways to provide electrical services in a carbon-constrained world? Our society built coal-fired power plants by counting cost but not carbon. Nuclear advocates defend their preference by counting carbon but not cost. But to protect the climate, we must save the most carbon at the least cost and in the least time, counting all three variables—carbon and cost and time. Costly options save less carbon per dollar than cheaper options. Slow options save less carbon per year than faster options. Thus even a low- or no-carbon option that is too costly or too slow will reduce and retard achievable climate protection. Being carbon-free does not establish climate-effectiveness.
Most analysts ignore common-sense comparisons of both cost and speed. The result is akin to arguing that since people are hungry, hunger is urgent, and caviar and rice are both food, therefore both are vital to reducing hunger. Since in reality money and time are both limited, our priorities in feeding people or in providing energy services must be informed by relative cost and speed. Lower cost saves more carbon per dollar. Faster deployment saves more carbon per year. We need both.
The bedrock economic principle of “opportunity cost” means you can’t spend the same money on two different things at the same time. Each purchase foregoes others. Buying nuclear power displaces buying some mixture of fossil-fueled generation, renewable generation, and efficient use. Nuclear owners strive to beat coal and gas while their allies often disparage or suppress renewables. Yet most US nuclear plants are uneconomic just to run, so many are closing. To keep milking those old assets instead, their powerful owners seek and often get multi-billion-dollar bailouts from malleable state legislatures for about a tenth of the nuclear fleet so far, postponing the economic reckoning by shooting the market messenger.
Such replacement of market choices with political logrolling distorts prices, crowds out competitors, slows innovation, reduces transparency, rewards undue influence, introduces bias, picks winners, invites corruption, and even threatens to destroy the competitive regional power markets where renewables and efficiency win. These are high prices for small or negative benefits (so far, displacing fossil-fueled or renewable generation, or savings, of about 2% of U.S. electricity).
Yet many political leaders think climate’s urgency demands every option, including preserving nuclear power at any cost. So what is that cost, construed in the narrowest economic terms?
Costs of new nuclear power vs. competing options
On 7 November 2019, the eminent 170-year-old financial house Lazard published its 13th annual snapshot of relative 2019-$ prices for different ways to generate a megawatt-hour of electricity. The analysis is authoritative though imperfect. Lazard removed renewables’ temporary and declining tax credits, but tacitly retained nonrenewables’ larger permanent subsidies—which for new US nuclear plants rival or exceed their construction cost and also subsidize nuclear operating cost more than, say, windpower gets. Lazard also favored fossil-fueled and nuclear power plants by assuming the same financing cost and structure for every kind of generator, though actually renewables win cheaper capital because of their faster construction and lower risk. And Lazard assumed nuclear operating costs below the actual average reported by the Nuclear Energy Institute, the industry’s lobbying organization. Nonetheless, Lazard’s comparison between new electricity resources is stark:
Nuclear power: $118–192/MWh (of which $29 is typical operating cost)
Coal power: $66–132/MWh (of which $33 is typical operating cost)
Combined-cycle gas power: $44–68/MWh
Utility-scale solar power: $32–42/MWh
Onshore windpower: $28–54/MWh
Previous editions also listed efficient use of electricity at $0–50/MWh, typically around $25/MWh. Efficiency, being already delivered to your meter, also avoids roughly $42/MWh of average delivery cost that all remote generators incur.
The other preeminent data sources—Bloomberg New Energy Finance, Lawrence Berkeley National Lab, National Renewable Energy Lab—show market prices for renewables broadly similar to Lazard’s, though BNEF says new nuclear power ranges from Lazard’s high value to nearly twice that big. Either way, new reactors are many-fold out of the money, with no material relief in prospect from any new technology, size, or fuel cycle. Nuclear costs are also rising while renewable costs keep falling. New nuclear plants will save many-fold less carbon per dollar than competing carbon-free resources, in proportion to their relative costs. And new reactors’ expected performance must be tempered by historical experience: of the 259 power reactors ordered in the US, by mid-2017 only 28 units or 11% had been built, were still competitive in their regional markets, and hadn’t suffered at least one outage lasting at least a year. That high dry-hole risk reinforces capital markets’ skepticism.
Should existing nuclear plants keep operating?
Today’s hot question, though, is not about new US reactors, which investors shun, but about the 96 existing reactors, already averaging about a decade beyond their nominal original design life. Most now cost more to run—including major repairs that trend upward with age—than their output can earn. They also cost more just to run than providing the same services by building and operating new renewables, or by using electricity more efficiently. So let’s go step by step through an eyechart about nuclear operating costs—which exclude original construction and financing costs (all sunk and usually amortized), but include those costs that need not be paid if the plant is closed.
The blue-green zone shows the range of US wholesale electricity prices for the past 15 years in constant 2014 dollars. The blue dots for windpower, and the brown dots for utility-scale solar photovoltaics, show the average fixed prices set in long-term private-market Power Purchase Agreements or PPAs. You can see that average new wind and solar power now sell at or below the lowest wholesale prices from nonrenewables, and trend downward, so windpower in 2018 is as low as $11/MWh or 1.1¢/kWh. That’s net of wind and solar power’s temporary federal subsidies, now phasing out, but they’re no longer important: the lower-right corner of the graph shows as diamonds the comparable unsubsidized prices of wind and solar in Chile, Mexico, and Morocco. The open squares at the far right are bids for Colorado solar and windpower; the filled squares add electricity storage with only modest cost. Actually, battery storage, though often cost-effective today, is rarely needed to “firm” the output of variable renewables (photovoltaics and windpower), because there are eight ample cheaper methods. Empirically, variable renewables’ firming costs are small, typically below $5/MWh, but appear to be considerably larger for central thermal power stations, which are allowed to ignore them as socialized system costs, whereas renewables are often expected to pay their own firming costs.
Now compare US nuclear plants’ average operating cost excluding all original construction cost. Average nuclear operations, the small magenta triangles, now cost more than new modern renewables, with or without their temporary subsidies. During two recent three-year periods, the magenta horizontal bars reported by the Nuclear Energy Institute show that those average nuclear operating costs by quartile fell as the worst reactors were closed—but renewable prices fell even faster. Nuclear operating costs will be hard to cut much further in reactors averaging four decades old, but renewable prices promise strong further declines for decades to come. International nuclear operating costs tend to be even higher.
These operating-cost data reveal an important climate opportunity. Operating cost (opex) exceeds $40/MWh for the costlier-to-run half of the 96 US power reactors, or $50 for the costliest quartile. Owners demand big new subsidies to keep running these money-losing plants. Yet customer efficiency costs utilities only $20–30/MWh on average—less if they shop carefully. Therefore closing a top-quartile-cost nuclear plant and buying efficiency instead, as utilities could volunteer or regulators require, would save considerably more carbon than continuing to run the nuclear plant. Some modern renewables too can now rival efficiency’s cost and could compete for that opportunity.
Thus, while we close coal plants to save carbon directly, we should also close distressed nuclear plants and reinvest their large saved operating cost in cheaper options to save carbon indirectly. These two climate-protecting steps are not alternatives; they are complements.
Replacing a closed nuclear plant with efficiency or renewables empirically takes only 1–3 years. If owners don’t give such advance notice—a common tactic to extort subsidies by making closure more disruptive—more natural gas might temporarily be burned, but then more than offset over the following years by the carbon-free substitutes. California’s biggest utility will therefore replace its well-running Diablo Canyon reactors with least-cost carbon-free resources to save money and carbon and to help the grid work better.
To get these outcomes, we must track not just carbon but also money and time. Investing judiciously, not indiscriminately, saves the most carbon per dollar. What about per year?
Which technologies are faster to deploy?
Claims that nuclear generation grows faster than renewable generation, making nuclear power desirable or even essential for climate protection, use cherry-picked old data and a strange methodology based on not absolute but per-capita growth. This makes the climate importance of decarbonizing power plants depend on the population of the country where it occurs, so Sweden or Slovakia look far more important than China. A 2016 paper with some distinguished coauthors in the leading journal Science used that methodology—more suited to comparing countries than technologies—to claim in this widely republished graph that nuclear growth is generically “much faster” than renewable growth:
But once I’d corrected that paper’s seven analytic flaws and distortions, the opposite finding emerged, despite using the same flawed per-capita methodology and the same data source: nuclear and renewable output can actually grow at similar speeds, and adding the three latest years of data reveals that renewables are pulling ahead, as the following graph shows. If, unlike the Science authors, we compare nuclear in ten countries with renewable growth in the same ten countries, renewables are faster in seven. But rapid nuclear growth occurred over three decades ago under conditions that no longer exist, while comparable or faster renewable growth is here, now, and accelerating. (In two years, China and India have just both increased their renewable generation by more than all their nuclear plants produced in 2018.)
Interestingly, all nine of the Science authors’ exemplary nuclear programs—not only in the United States—are in trouble. Sweden’s booming windpower will pass its capped nuclear capacity this year; France cannot afford to fix or replace its troubled nuclear fleet; Belgium’s capped nuclear capacity is unreliable and safety-challenged; Slovakia’s newbuild project is hobbled by delay, cost, and corruption; Taiwan, South Korea, and Germany have nuclear phase-out policies; and in Japan, most reactors remain shut down eight years after Fukushima, and their lost output plus energy to support Japan’s 11% GDP growth has already been 97% offset by savings and renewables.
The graphed speed comparisons reflect the technologies’ basic attributes. Nuclear plants take many years to build, typically around a decade, while renewable projects can take a year or less—even months or weeks. Further, national nuclear power programs need three times as much lead time for institutional preparations as modern renewables need. For both reasons, renewables can start saving carbon many years sooner. Both project-level and program-level nuclear slowness incur a big carbon penalty. Being both slower and costlier than its modern competitors makes nuclear power doubly unhelpful for protecting the climate.
Yet undaunted by actual data, propagandists keep spreading the myth of nuclear speed. The 2016 Science paper, though discredited by three devastating journal critiques, was the sole basis cited (“several studies” were mentioned but no others cited) by the World Nuclear Association’s 24 September 2019 press release “Combatting climate change faster with new nuclear build.” A day later, with no citation, WNA trumpeted: “Nuclear energy is the fast track to decarbonization.” These headlines have no basis in fact. WNA seems as challenged by math as by truth: it cannot even correctly describe what the previous graph shows.
The evidence about relative speed becomes simpler and stronger if we simply compare different power sources’ total output growth worldwide, requiring no international comparisons that mix countries with technologies. Worldwide, you can see that the blue curve, modern renewables, grew faster in the past decade than the magenta nuclear curve’s fastest growth, which occurred more than three decades ago.
Total renewables (green), which include big hydro, already outgenerate natural gas. Modern renewables, which omit big hydro, passed nuclear generation in 2016, quietly passed a trillion watts—a terawatt (TW)—of installed capacity in 2017, and are accelerating. That first TW took ~15 years; the next TW is set to take 4–5 years and cost about half as much, with most of the renewables made and nearly half bought by China. In 2013, China added more PV capacity than the US had added in total since developing it 59 years earlier. In 2016, China doubled the pace to three football fields per hour, including 11 GW in June. In 2018, China added more PVs in June than the US added all year. The world’s installed solar capacity is now >50⨉ IEA’s 2002 forecast. In 2028—sooner than building a typical nuclear plant—Bloomberg predicts solar electricity alone will exceed 2018 global nuclear generation in 2018, and even sooner, solar power is set to add 1 TW each year.
Global carbon-free electricity is now less than one-third nuclear. Counting also carbon-free production of non-electric energy—biofuels and modern renewable heat—nuclear power struggles to sustain less than one-fourth of the world’s carbon-free final energy use. Why pay more to revive it at the expense of faster and cheaper competitors? Sustaining uneconomic reactors would not only divert public funding from more climate-effective competitors but also constrain their sales and degrade the competitive markets where they thrive. Slowing and blocking the fastest and cheapest climate solutions harms climate protection.
Why should a particular low-carbon solution, unable to compete after a half-century, now be awarded walled-garden market preferences (blocks of demand mandatorily served by nuclear power and not contestable by competitors) and new subsidies unavailable to other low-carbon solutions? Shouldn’t policy, as IEA agrees, be technology-neutral?
How high can US nuclear subsidies go?
Meanwhile, back in the United States, the climate-effectiveness of continued nuclear operations is not discussed; the conversation focuses solely on carbon, not on cost or time. Indeed, the industry’s immense lobbying power has now hatched a brazen new way to make taxpayers or customers pay for existing nuclear plants for a seventh time, and disadvantage their most potent supply-side competitor (modern renewable power), and reduce and retard climate protection while claiming to increase it. Rarely have so many been so deceived so thoroughly, for so long, at such cost.
The previous six forms of payment were:
1. Taxpayers paid to create the nuclear industry, build its fueling infrastructure, and finance the reactor fleet via a vast array of often opaque and generally permanent federal subsidies that cost more than building the plants and more than the value of their output.
2. Through generally regulated utility tariffs, customers paid for the plants’ construction and financing, including a just and reasonable return on capital.
3. Customers paid over decades for the plants’ operation, including major repairs, power upratings, and safety upgrades.
4. Many customers reimbursed owners for “stranded-asset costs” totaling upwards of $70 billion to support the owner-demanded transition to competitive wholesale markets.
5. Over the past few years, when reactors generating 2 percent of U.S. electricity proved unable to compete in those wholesale markets (though most of their owners kept their finances secret and kept reporting profits to investors), the owners persuaded state legislators in Illinois, New York, New Jersey, Connecticut, and Ohio to vote billions of dollars a year for new multi-year operating subsidies.
6. Exelon, the nation’s largest nuclear operator and the leading player in the previous two steps, successfully sought Federal regulatory approval for greater capacity payments from power pools whose auctions found nuclear power uncompetitive and whose own rules were thrown off-balance by the new state subsidies. And now, as the annual logrolling season of “tax extenders” rolls around in Congress:
7. For the third year, Exelon is advancing a “Nuclear Powers America Act” to create a new federal investment tax credit on nuclear fuel and maintenance expenses to “help level the playing fuel with other clean energy sources”—whose temporary tax credits are meanwhile being phased down or out. This follows a longstanding pattern of giving different kinds of subsidies to renewables than to nuclear, then “leveling the playing field” by trying to duplicate renewables’ specific forms of subsidies with new ones for nuclear, but never the reverse.
This saga of selling the same hay seven times—and those clever lawyers aren’t done yet—doesn’t include many additional federal and state subsidies. It also doesn’t include an emerging potential scam that could end up with ratepayers’ and federal taxpayers’ getting stuck with vast decommissioning and waste-management liabilities. This emerging pattern has an LLC buy a closed reactor. Absent oversight or rules to stop misbehavior, the LLC could then choose to strip the accumulated customer-funded multi-billion-dollar cash decommissioning fund, not finish the job, and walk away, leaving the parent company whole and electricity customers or taxpayers holding the bag. Watch this space.
Exelon’s proposed “nuclear investment tax credit” has ingenious new features:
– By redefining normal accounting categories so fuel becomes a capital investment, it repays utilities for an “investment” that’s really just a normal operating cost—thus trying to make nuclear operating costs look small by shifting much of them to taxpayers.
– The nuclear operating costs it covers have no counterpart for renewables (fuel, nuclear waste management, protection against catastrophic releases of radioactivity), or almost none (operation & maintenance costs), so a tax credit for them would specifically advantage nuclear against renewables.
– Nuclear owners may be able to double-dip, collecting the new federal subsidy and new state subsidies for the same plants and thus turning dead dinosaurs into juicy cash cows.
– There’s virtually no “means test”: the new federal subsidy would apply to about 95% of US operating reactors, including those that the industry claims are currently profitable.
– The proposed legislation, obscurely written in tax-law jargon, appears to be a 30% tax credit (phasing down to 26% in 2024, 22% in 2025, and a permanent 10% in and after 2026), and to cost ~$22–26 billion over the first decade, or ~$33 billion counting the crowding-out of cheaper competitors. Every billion dollars thus bilked from taxpayers is unavailable to provide more electrical services and save more carbon by cheaper means.
– By further distorting the delicate balance between federal, regional, and state regulation, the subsidy seems tailored to weaken or destroy the efficient regional power markets where renewables beat nuclear power. The goal is thus to pay nuclear power for values it doesn’t deliver, while blocking its most potent competitors from continuing to provide the values they do deliver.
– Unlike renewable credits that have helped to mature important new technologies, the nuclear credit would elicit no new production, capacity, or innovation. It would simply transfer tens of billions of dollars to the owners of uncompetitive nuclear assets bought decades ago—if they apply for license extension by 2026, as nearly all have done.
This covert attack on renewables is logical because renewables are now the supply-side competitor nuclear must beat but can’t. The nuclear industry is reluctant to admit that renewables are a legitimate competitor, since this would contradict its claims that renewables can’t supply reliable power. Renewables, unlike nuclear power, are also widely popular. Nuclear advocates therefore tend to blame their woes instead on cheap natural gas. However, new and often even existing combined-cycle gas-fired power plants no longer have a business case: a September 2019 study found that at least 90% of the 88 proposed US gas-fired plants are pre-stranded assets. Oddly, the nuclear industry has also ignored a suggestion that would advantage nuclear power (but also renewables and efficiency) against natural gas by properly counting the market value of its price volatility. (Basic financial economics requires such risk-adjustment when comparing a volatile cost stream with a fixed one, such as efficiency and renewables incur.)
While some nuclear advocates favor carbon pricing (which would comparably advantage nuclear power, renewables, and efficiency over gas-fired power), many do not, presumably because the same utilities often run both nuclear and fossil-fueled plants. Nuclear advocates must awkwardly straddle the need to oppose fossil-fueled generation when seeking supposedly “green” state subsidies while cozying up to fossil-fueled generation at the federal level, where the Secretary of Energy wants to bail out both nuclear and coal plants.
In sum, the nuclear industry seeks its own sales arrangements protected from competition, its own prices determined by political processes rather than markets, and diminished opportunities for its carbon-free competitors to express their value, reach their customers, and discover their own prices. This could be good for compliant legislators’ campaign contributions, but hardly in the national interest or helpful for climate protection.
Such anti-market monkeybusiness cannot indefinitely forestall the victory of cheaper competitors, but it can delay and diminish climate protection while transferring tens of billions of unearned dollars from taxpayers and customers to nuclear owners. That would save less carbon, more slowly, than letting the best buys win, yet some politicians fervently favoring climate protection mistakenly endorse it, and most journalists reinforce their misconception.
Citizens who care about climate or markets or both should therefore pay attention not only to carbon but not also to cost and time—and should vigorously defend markets’ ability to choose climate solutions that can save the most carbon per dollar and per year. Doing otherwise makes climate change worse than it could and should have been, stifles economic competitiveness and innovation, and suppresses market competition. That these errors are often bipartisan does not excuse them. Our best climate strategy would be to start taking economics seriously