Nuclear power has suffered the greatest collapse of any enterprise in the industrial history of the world. The twentieth century ended with installed nuclear capacity less than 10%, and an ordering rate less than 1%, of the lowest IAEA forecasts made a quarter-century ago. No vendor has made money by selling reactors, though some have made it up on repairs. Worldwide, nuclear power is stuck at 11% of total generating capacity, providing 6.3% of 1998 primary energy output, vs. 8.8% for renewables without, or about 20.3% with, traditional biofuels. In the U.S., nuclear investments exceeding a trillion dollars are delivering little more energy than biofuels; nuclear power’s primary energy output and installed capacity are roughly the same as renewables’, though its electric output is about four-fifths higher due to nuclear plants’ recently improved capacity factors.

The basic reason for this market disappointment is not public concern but unfavorable economics. As The Economist’s 19 May 2001 cover story concludes, nuclear power has gone from too cheap to meter to too costly to matter, especially in an increasingly competitive marketplace. A nuclear plant with zero capital cost is now cheaper to write off (and give away equivalent electricity-saving equipment instead) than to operate. Nuclear power’s unpleasant capital- and repair-cost escalation and technical surprises are worldwide phenomena, independent of politics. Even France’s nuclear program was outpaced twofold by energy efficiency, bankrupted its operator in all but name, and is unlikely to be replaced by more reactors. During 1990—99, global capacity growth averaged only 1%/y for nuclear power (to 354 GW), vs. 17%/y for photovoltaics (PVs) and 24%/y for windpower. That’s no longer simply rapid growth from a tiny base: in the 1990s, nuclear power averaged global additions of 3.1 GW/y, yet in 2000—01, windpower is adding 3.5—5 GW/y, and by the end of 2001 should reach 22 GW.

Nuclear power’s main if not sole competitor was originally presumed to be giant coal plants. These are now equally obsolete on the margin: many competitors work better and cost less than either. That’s why central thermal plants are seldom ordered nowadays except by a handful of centrally planned energy systems. Nuclear salesmen scour the world for a single order, while vendors of combined-cycle gas plants, microturbines, wind turbines, PVs, and energy efficiency strive to meet bulging order books.

If no existing nuclear plant suffers an accidental or malicious radioactive release serious enough to compel their early shutdown, they’ll probably operate until they’re too costly to maintain. In theory, they must also compete in economic dispatch (lowest short-run marginal cost of operation), but in fact, many U.S. reactors have won "must-run" status entitling them to run whenever available even if uneconomic, which ~20—25% of them already are against market-clearing prices. (In Sweden, Sydkraft unsuccessfully sought compensation for not continuing to lose money by running Barsebäck at an operating cost roughly twice the price of imported power.) The U.S. short-run marginal cost of nuclear electricity at the busbar is typically reckoned at ~$0.015—0.039/kWh; the average in 2000, under the industry’s restrictive definition of costs, was about $0.018. (This article uses constant year-2000 U.S. dollars throughout. Net capital additions–major repairs that are really hidden operating costs–increase the industry’s declared operating costs, as would fully internalized waste, decommissioning, and major-accident costs, now typically socialized.)

In any event, the cost of delivering the power to the customer must also be added for fair comparison with onsite options that require no delivery. Delivering the average U.S. kWh costs roughly $0.025/kWh for the capital and operating costs and the losses embedded in the existing grid; residential delivery costs, or costs of marginal delivery capacity (new transmission and distribution capacity), are often several-fold higher, but let’s use the historic average figure as a conservatism. Nuclear power’s typical short-run marginal cost, delivered to the customer, is then about $0.045/kWh using the industry’s narrow cost definition. But its long-run marginal delivered cost (including building a new nuclear plant) is at least $0.10—0.15/kWh. (Enthusiasts claim that building and running a hypothetical South African pebble-bed design might cost perhaps $0.05/kWh delivered, but given historic experience with paper vs. real reactors, twice that would be more plausible.) Existing coal plants often run at or below $0.02 and hence deliver at about $0.045/kWh for existing plants; cleaner new ones would raise that to roughly $0.06—0.08/kWh, almost certainly cheaper than new nuclear plants. What else costs less to build and run than nuclear power?

The most potent competitor, many-fold cheaper than just running existing nuclear plants anywhere in the world, is end-use efficiency–a resource that’s also many-fold larger. Its average historic U.S. cost, including many poorly designed and chiefly residential programs, is only $0.02 per saved kW-h, already delivered to the customer’s premises. Well-designed retrofits, however, can save most of the electricity now used at empirical costs typically ~$0.005/kWh in the business sectors (a 16-month payback at a 5-cent tariff)–even less than zero in new buildings and factories. (My own 90% saving in household electricity had a 10-month payback in 1984; today’s technologies are far better and cheaper.) Saved energy is currently the United States’ largest single source of energy services, delivering two-fifths of their total. Reduced primary energy intensity during 1975—2000 is now an energy "source" over five times as big as domestic oil output, over twice total U.S. oil imports, over 12 times Persian Gulf imports, and 7.7 times primary nuclear energy output. Oil productivity has indeed doubled since 1975.

Electrical productivity is far less mature, having improved only 8% during 1975—2000–partly because electricity has been heavily subsidized and promoted, has typically been priced at average cost rather than on the margin, and is distributed by utilities that in 49 states are rewarded for selling more energy and penalized for cutting customers’ bills. Nonetheless, electrical productivity since 1997 has sustained its steadiest gains in history, averaging 1.6% per year–half as fast as the aggregate drop in primary energy intensity. One-fourth of the national electrical saving has come from California, even after its demand-side efforts were derailed in the mid-1990s. California has held its per-capita use of electricity nearly flat for the past quarter-century, saving 25 average GW compared with the intensity prevailing in the rest of the U.S., and boosting its economic output by investing more productively the billions of dollars thereby saved. Analogously, the Electric Power Research Institute, at its 2001 summer seminar, presented an "enhanced productivity" scenario envisaging 3%/y gains in electrical productivity (a 45% drop in intensity) during 2000—20.

After end-use efficiency, typically the next cheapest competitors are three kinds of new generators:

• combined-cycle gas turbines (often ~$0.05—0.06 per delivered kWh at 1999 prices, including 30-year fixed-price gas contracts, or about a penny more at 2001 prices due to temporary turbine and gas shortages),
• some renewables (notably, well-sited windpower costs ~$0.055—0.06 per delivered kWh and will soon drop to or below $0.05, minus a $0.015 subsidy in the U.S.), and

• onsite and hence requiring no delivery, industrial and commercial co- and trigeneration from larger units or from microturbines, again using 30-year constant-price gas contracts (~$0.005—0.05, often ~$0.01—0.02, per delivered kWh net of thermal credit).

In short, any of three abundant and readily purchasable resources–efficient end-use, efficiently used gas (especially when thermally integrated), and windpower–can easily beat new and can even beat most old nuclear plants on private internal cost. Any one of these three competitive resources would make nuclear power unnecessary and uneconomic. Two of them are climate-safe, and the third has very low climate impact. Even though cogeneration is typically gas-fired, full practical and profitable conversion to it would cut U.S. CO₂ emissions by about 23%.

As IIASA first showed for inefficient solar trough systems two decades ago, renewable power generation is no more land-intensive than the full nuclear or coal fuel cycle. Denmark, now 16% wind-powered, is on target for 50% by 2030, without issues of land-use or intermittence (an old canard long since resolved by wind diversity on a mesoscale). A fifth of U.S. electricity could be made by modern wind turbines occupying 5% of four Montana counties, or 0.6% of the Lower 48 states; or all of U.S. annual electricity could come from ordinary PVs occupying half of a square 160 km on a side (though one would actually site them in distributed fashion, chiefly on buildings). In either case, the land-use can be shared, much as farmers and ranchers do now, earning valuable revenues from the wind that blows above their herds and crops. Modern renewable sources repay their energy investments in months to a few years: illustrating their materials-frugality, 1 kg of silicon in thin-film PVs can produce more electricity than 1 kg of uranium in a pressurized-water reactor. Further PV breakthroughs continue (e.g., Science 293:1119—1122 (2001)).

Meanwhile, the next revolution–distributed utilities–is gaining momentum. As The Economist remarked (17 May 2001, online ed.), "In fact, the trend since the mid-1970s has been towards smaller plants. It is micropower, not megapower, that the market favours, thanks to the far smaller financial risk involved." My forthcoming book Small Is Profitable: The Hidden Economic Benefits of Making Electrical Resources the Right Size shows how roughly 125 "distributed benefits" of decentralized electricity sources can typically increase their economic value by roughly tenfold, making even PVs cost-effective today in most applications. Even more valuable than familiar electrical engineering attributes–lower grid costs and losses, higher reliability and power quality, dispatchable reactive power, near-infinite ramp rates, etc.–are the lower financial risks of small, short-lead-time modules. For example, a 10-kW resource installable overnight could cost 2.7 times as much per kW as a 50-MW resource installable in two years, yet yield identical financial performance. In the new arena now emerging, discount rates are project-specific and risk-adjusted, new market actors understand financial economics, and competition increasingly embraces all options. Continuing nuclear ownership consolidation may improve operations more than it concentrates risks, but the firesale prices realized at used-reactor sales in the late 1990s confirm a market perception of low or even negative long-run asset value.

Nuclear waste accumulates, and neutron fluence raises decommissioning costs, proportionally to nuclear power generation. Moreover, nuclear power’s having died of an incurable attack of market forces–plus the end of the Cold War–offers a unique opportunity to inhibit proliferation as fissile materials, skills, and equipment become no longer ordinary items of commerce (except for minor and readily safeguarded medical and industrial uses). This would make such bomb-kit ingredients harder to get, more conspicuous to try to get, and politically far costlier to be caught trying to get or supply, because for the first time, the reason for wanting them would be unambiguously military. This exposure of illicit transactions–now hidden in and rationalized by a vast flow of supposedly innocent civilian nuclear commerce–would not make nuclear bomb proliferation impossible, but would make it far more difficult, and would focus its resource flows into narrower, more readily monitored channels. This potential for an internally consistent nonproliferation policy was summarized in Foreign Affairs in Summer 1980 and in the 1979 book Energy/War: Breaking the Nuclear Link; today, its preconditions, somewhat visionary at that time, have been realized, and the logic is as tight as ever.

Nuclear power is unrelated to this problem save in the fundamental sense that its high "stranded asset" costs helped trigger the dreadful restructuring, and in the minor sense that a 1.1-GW outage at San Onofre contributed to the problem. More nuclear power certainly wouldn’t help timely (or economically) even it could be financed and sited. In fact, as Barron’s 6 August 2001 cover story noted, the U.S. is already building more power plants than it might plausibly need. Such overshoot, last seen in the mid-1980s, occurs when slow-to-build central power plants collide head-on with quickly captured efficiency. In the first six months of 2001, for example, even before price hikes hit, Californians’ electric savings undid the previous 5—10 years’ demand growth, helping to stabilize the market and frustrate the ticket-scalpers’ ambitions.

Nuclear advocates’ last hope is that climate concerns will revitalize their option. Alas, they’ve overlooked opportunity cost–the impossibility of spending the same money on two different things at the same time. If saving a kW-h cost (pessimistically) as much as three cents, while delivering a kW-h of new nuclear electricity cost (very optimistically) as little as six cents, then the six cents spent for each new nuclear kW-h could instead have bought two kW-h worth of efficiency. The nuclear purchase therefore displaced one less kW-h of coal-fired electricity than the same money could have done by buying the cheaper (efficiency) option instead. That’s why the order of economic priority must also be the order of environmental priority; why it’s irrelevant whether nuclear power can beat coal power as long as any other option costs still less; and why nuclear power makes global warming worse.

Nuclear power is a future technology whose time has passed. Its economic problems are so ineluctable that it would still fail even if it had no political, environmental, safety, or security problems. And as the retirement of the older, higher-quality nuclear pioneers continues the worrisome trend predicted by Nobel physicist Hannes Alfvén–that the enterprise will "pass into ever less competent hands"–each dollar spent to address the unsolved problems will buy ever less solution.

Looking after nuclear power’s legacy will be a business with a long future; but the operation of today’s nuclear fleet will not. Despite immense investments, devoted efforts, and dedicated careers, those plants will long stand as a monument to what happens when a technology avoids market and political accountability for long enough to make really big mistakes, and when its advocates develop a reputation for mendacity. Its epitaph could be: "Here lies a technology that failed because it did not take its discipline from the marketplace, its values from its customers, and its design from nature." Its seemingly great promise was betrayed by tragic flaws.

Nuclear power has been called "a fit technology for a wise, farseeing, and incorruptible people." A pity we haven’t more of them. But its best legacy would be not to make the same mistake again; and the best way to do that is to take economics seriously. The Bush Administration’s claims that nuclear power is safe, but needs an extension of its unique statutory cap on liability for major accidents, and is economic, but needs another $1.5 billion in tax breaks, is perplexing to advocates of free markets, and hardly seems a propitious start at acknowledging and respecting market outcomes.

Shorn of distracting details, the nuclear power issue is simple. The technology has failed in the marketplace–a tragic misallocation of talent, work, hope, and investment that deserved better and that continues to distort public choices. However, accepting the verdict of the marketplace will yield the right energy policy conclusion and will also simplify the politics of finding the least unsatisfactory place to put nuclear waste. An orderly terminal phase should be designed for this unfortunate mistake–but can’t be, so long as nuclear theology dominates policy. If recognized, however, the commercial collapse of nuclear power, and the rise of better energy alternatives, could be turned into the long-awaited missing step toward effective nonproliferation.

* * *