By Amory B. Lovins, Imran Sheikh, and Alex Markevich
Rocky Mountain Institute
April 28, 2008
Nuclear power, we’re told, is a vibrant industry that’s dramatically
reviving because it’s proven, necessary, competitive, reliable, safe,
secure, widely used, increasingly popular, and carbon-free—a perfect
replacement for carbon-spewing coal power. New nuclear plants thus
sound vital for climate protection, energy security, and powering a
There’s a catch, though: the private capitalmarket isn’t investing
in new nuclear plants, and without financing, capitalist utilities
aren’t buying. The few purchases, nearly all in Asia, are all made by
central planners with a draw on the public purse. In the United States,
even government subsidies approaching or exceeding new nuclear power’s
total cost have failed to entice Wall Street.
This non-technical summary article compares the cost, climate
protection potential, reliability, financial risk, market success,
deployment speed, and energy contribution of new nuclear power with
those of its low- or no-carbon competitors. It explains why soaring
taxpayer subsidies aren’t attracting investors. Capitalists instead
favor climate-protecting competitors with less cost, construction time,
and financial risk. The nuclear industry claims it has no serious
rivals, let alone those competitors—which, however, already outproduce
nuclear power worldwide and are growing enormously faster.
Most remarkably, comparing all options’ ability to protect the
earth’s climate and enhance energy security reveals why nuclear power
could never deliver these promised benefits even if it could find
free-market buyers—while its carbon-free rivals, which won $71 billion
of private investment in 2007 alone, do offer highly effective climate
and security solutions, sooner, with greater confidence.
observed in 2001 that “Nuclear power, once claimed to be too cheap to
meter, is now too costly to matter”—cheap to run but very expensive to
build. Since then, it’s become several-fold costlier to build, and in a
few years, as old fuel contracts expire, it is expected to become
several-fold costlier to run. Its total cost now markedly exceeds that
of other common power plants (coal, gas, big wind farms), let alone the
even cheaper competitors described below.
Construction costs worldwide have risen far faster for nuclear than
non-nuclear plants, due not just to sharply higher steel, copper,
nickel, and cement prices but also to an atrophied global
infrastructure for making, building, managing, and operating reactors.
The industry’s flagship Finnish project, led by France’s top builder,
after 28 months’ construction had gone at least 24 months behind
schedule and $2 billion over budget.
By 2007, as Figure 1 shows, nuclear was the costliest option among
all main competitors, whether using MIT’s authoritative but now low
2003 cost assessment1, the Keystone Center’s mid-2007 update
(see Figure 1, pink bar), or later and even higher industry estimates
(see Figure 1, pink arrow)2.
Cogeneration and efficiency are “distributed resources,” located
near where energy is used. Therefore, they don’t incur the capital
costs and energy losses of the electric grid, which links large power
plants and remote wind farms to customers3. Wind farms, like solar cells4,
also require “firming” to steady their variable output, and all types
of generators require some backup for when they inevitably break. The
graph reflects these costs.
Making electricity from fuel creates large amounts of byproduct
heat that’s normally wasted. Combined-cycle industrial cogeneration and
buildingscale cogeneration recover most of that heat and use it to
displace the need for separate boilers to heat the industrial process
or the building, thus creating the economic “credit” shown in Figure 1.
Cogenerating electricity and some useful heat from currently discarded
industrial heat is even cheaper because no additional fuel is needed5.
End-use efficiency lets customers wring more service from each
kilowatthour by using smarter technologies. As RMI’s work with many
leading firms has demonstrated, efficiency provides the same or better
services with less carbon, less operating cost, and often less up-front
investment. The investment required to save a kilowatt-hour averages
about two cents nationwide, but has been less than one cent in hundreds
of utility programs (mainly for businesses), and can even be less than
zero in new buildings and factories—and in some retrofits that are
coordinated with routine renovations.
Wind, cogeneration, and end-use efficiency already provide
electrical services more cheaply than central thermal power plants,
whether nuclear- or fossil-fuelled. This cost gap will only widen,
since central thermal power plants are largely mature while their
competitors continue to improve rapidly. The high costs of conventional
fossil-fuelled plants would go even higher if their large carbon
emissions had to be captured.
Uncompetitive CO2 Displacement
Nuclear plant operations emit almost no carbon—just a little to produce the fuel under current conditions6.
Nuclear power is therefore touted as the key replacement for coal-fired
power plants. But this seemingly straightforward substitution could
instead be done using non-nuclear technologies that are cheaper and
faster, so they yield more climate solution per dollar and per year. As
Figure 2 shows, various options emit widely differing quantities of CO2
per delivered kilowatt-hour.
Coal is by far the most carbonintensive source of electricity, so
displacing it is the yardstick of carbon displacement’s effectiveness.
A kilowatthour of nuclear power does displace nearly all the 0.9-plus
kilograms of CO2 emitted by producing a kilowatt-hour from coal. But so
does a kilowatthour from wind, a kilowatt-hour from recovered-heat
industrial cogeneration, or a kilowatt-hour saved by end-use
efficiency. And all of these three carbonfree resources cost at least
one-third less than nuclear power per kilowatt-hour, so they save more
carbon per dollar.
Combined-cycle industrial cogeneration and building-scale
cogeneration typically burn natural gas, which does emit carbon (though
half as much as coal), so they displace somewhat less net carbon than
nuclear power could: around 0.7 kilograms of CO2 per kilowatt-hour7.
Even though cogeneration displaces less carbon than nuclear does per
kilowatt-hour, it displaces more carbon than nuclear does per dollar
spent on delivered electricity, because it costs far less. With a net
delivered cost per kilowatthour approximately half of nuclear’s,
cogeneration delivers twice as many kilowatt-hours per dollar, and
therefore displaces around 1.4 kilograms of CO2 for the same cost as
displacing 0.9 kilograms of CO2 with nuclear power.
Figure 3 compares different electricity options’ cost-effectiveness
in reducing CO2 emissions. It counts both their cost-effectiveness, in
delivering kilowatthours per dollar, and their carbon emissions, if any.
Nuclear power, being the costliest option, delivers less electrical
service per dollar than its rivals, so, not surprisingly, it’s also a
climate protection loser, surpassing in carbon emissions displaced per
dollar only centralized, non-cogenerating combined-cycle power plants
burning natural gas8. Firmed windpower and cogeneration are
1.5 times more costeffective than nuclear at displacing CO2. So is
efficiency at even an almost unheard-of seven cents per kilowatthour.
Efficiency at normally observed costs beats nuclear by a wide margin—
for example, by about ten-fold for efficiency costing one cent per
New nuclear power is so costly that shifting a dollar of spending
from nuclear to efficiency protects the climate several-fold more than
shifting a dollar of spending from coal to nuclear. Indeed, under
plausible assumptions, spending a dollar on new nuclear power instead
of on efficient use of electricity has a worse climate effect than
spending that dollar on new coal power!
If we’re serious about addressing climate change, we must invest
resources wisely to expand and accelerate climate protection. Because
nuclear power is costly and slow to build, buying more of it rather
than of its cheaper, swifter rivals will instead reduce and retard
All sources of
electricity sometimes fail, differing only in why, how often, how much,
for how long, and how predictably. Even the most reliable giant power
plants are intermittent: they fail unexpectedly in billion-watt chunks,
often for long periods. Of all 132 U.S. nuclear plants built (52
percent of the 253 originally ordered), 21 percent were permanently and
prematurely closed due to reliability or cost problems, while another
27 percent have completely failed for a year or more at least once.
Even reliably operating nuclear plants must shut down, on average, for
39 days every 17 months for refueling and maintenance. To cope with
such intermittence in the operation of both nuclear and centralized
fossil-fuelled power plants, which typically fail about 8 percent of
the time, utilities must install a roughly 15 percent “reserve margin”
of extra capacity, some of which must be continuously fuelled, spinning
ready for instant use. Heavily nuclear-dependent regions are
particularly at risk because drought, a serious safety problem, or a
terrorist incident could close many plants simultaneously.
Nuclear plants have an additional disadvantage: for safety, they
must instantly shut down in a power failure, but for nuclear-physics
reasons, they can’t then be quickly restarted. During the August 2003
Northeast blackout, nine perfectly operating U.S. nuclear units had to
shut down. Twelve days of painfully slow restart later, their average
capacity loss had exceeded 50 percent. For the first three days, just
when they were most needed, their output was below 3 percent of normal.
The big transmission lines that highly concentrated nuclear plants
require are also vulnerable to lightning, ice storms, rifle bullets,
and other interruptions. The bigger our power plants and power lines
get, the more frequent and widespread regional blackouts will become.
Because 98–99 percent of power failures start in the grid, it’s more
reliable to bypass the grid by shifting to efficiently used, diverse,
dispersed resources sited at or near the customer. Also, a portfolio of
many smaller units is unlikely to fail all at once: its diversity makes
it especially reliable even if its individual units are not.
The sun doesn’t always shine on a given solar panel, nor does the
wind always spin a given turbine. Yet if properly firmed, both
windpower, whose global potential is 35 times world electricity use,
and solar energy, as much of which falls on the earth’s surface every
~70 minutes as humankind uses each year, can deliver reliable power
without significant cost for backup or storage. These variable
renewable resources become collectively reliable when diversified in
type and location and when integrated with three types of resources:
steady renewables (geothermal, small hydro, biomass, etc.), existing
fuelled plants, and customer demand response. Such integration uses
weather forecasting to predict the output of variable renewable
resources, just as utilities now forecast demand patterns and
hydropower output. In general, keeping power supplies reliable despite
large wind and solar fractions will require less backup or storage
capacity than utilities have already bought to manage big thermal
stations’ intermittence. The myth of renewable energy’s unreliability
has been debunked both by theory and by practical experience. For
example, three north German states in 2007 got upwards of 30% of their
electricity from windpower-39% in Schleswig-Holstein, whose goal is
100% by 2020.
Large Subsidies to Off set High Financial Risk
The latest U.S. nuclear plant proposed is estimated to cost $12–24
billion (for 2.2–3.0 billion watts), many times industry’s claims, and
off the chart in Figure 1 above. The utility’s owner, a large holding
company active in 27 states, has annual revenues of only $15 billion.
Such high, and highly uncertain, costs now make financing prohibitively
expensive for free-market nuclear plants in the half of the U.S. that
has restructured its electricity system, and prone to politically
challenging rate shock in the rest: a new nuclear kilowatt-hour
costing, say, 16 cents “levelized” over decades implies that the
utility must collect ~27 cents to fund its first year of operation.
Lacking investors, nuclear promoters have turned back to taxpayers,
who already bear most nuclear accident risks and have no meaningful say
in licensing. In the United States, taxpayers also insure operators
against legal or regulatory delays and have long subsidized existing
nuclear plants by ~1–5¢ per kilowatt-hour. In 2005, desperate for
orders, the politically potent nuclear industry got those subsidies
raised to ~5–9¢ per kilowatthour for new plants, or ~60–90 percent of
their entire projected power cost. Wall Street still demurred. In 2007,
the industry won relaxed government rules that made its 100 percent
loan guarantees (for 80 percent-debt financing) even more
valuable—worth, one utility’s data revealed, about $13 billion for a
single new plant. But rising costs had meanwhile made the $4 billion of
new 2005 loan guarantees scarcely sufficient for a single reactor, so
Congress raised taxpayers’ guarantees to $18.5 billion. Congress will
be asked for another $30+ billion in loan guarantees in 2008.
Meanwhile, the nonpartisan Congressional Budget Office has concluded
that defaults are likely.
Wall Street is ever more skeptical that nuclear power is as
robustly competitive as claimed. Starting with Warren Buffet, who just
abandoned a nuclear project because “it does not make economic sense,”
the smart money is heading for the exits. The Nuclear Energy Institute
is therefore trying to damp down the rosy expectations it created. It
now says U.S. nuclear orders will come not in a tidal wave but in two
little ripples—a mere 5–8 units coming online in 2015–16, then more if
those are on time and within budget. Even that sounds dubious, as many
senior energyindustry figures privately agree. In today’s capital
market, governments can have only about as many nuclear plants as they
can force taxpayers to buy.
The Micropower Revolution
While nuclear power
struggles in vain to attract private capital, investors have switched
to cheaper, faster, less risky alternatives that The Economist
calls “micropower”—distributed turbines and generators in factories or
buildings (usually cogenerating useful heat), and all renewable sources
of electricity except big hydro dams (those over ten megawatts). These
alternatives surpassed nuclear’s global capacity in 2002 and its
electric output in 2006. Nuclear power now accounts for about 2 percent
of worldwide electric capacity additions, vs. 28 percent for micropower
(2004– 07 average) and probably more in 2007–08.
An even cheaper competitor is enduse efficiency
(“negawatts”)—saving electricity by using it more effi ciently or at
smarter times. Despite subsidies generally smaller than nuclear’s, and
many barriers to fair market entry and competition, negawatts and
micropower have lately turned in a stunning global market performance.
Micropower’s actual and industry-projected electricity production is
running away from nuclear’s, not even counting the roughly comparable
additional growth in negawatts, nor any fossil-fuelled generators under
a megawatt (see Figure 4)9.
The nuclear industry nonetheless claims its only serious
competitors are big coal and gas plants. But the marketplace has
already abandoned that outmoded battleground for two others: central
thermal plants vs. micropower, and megawatts vs. negawatts. For
example, the U.S. added more windpower capacity in 2007 than it added
coal-fired capacity in the past five years combined. By beating all
central thermal plants, micropower and negawatts together provide about
half the world’s new electrical services. Micropower alone now provides
a sixth of the world’s electricity, and from a sixth to more than half
of all electricity in twelve industrial countries (the U.S. lags with 6
In this broader competitive landscape, high carbon prices or taxes
can’t save nuclear power from its fate. If nuclear did compete only
with coal, then far above- market carbon prices might save it; but coal
isn’t the competitor to beat. Higher carbon prices will advantage all
other zero-carbon resources—renewables, recoveredheat cogeneration, and
negawatts—as much as nuclear, and will partly advantage fossil-fueled
but low-carbon cogeneration as well.
Small Is Fast, Low-Risk, and High in Total Potential
Small, quickly built units are faster to deploy for a given total
effect than a few big, slowly built units. Widely accessible choices
that sell like cellphones and PCs can add up to more, sooner, than
ponderous plants that get built like cathedrals. And small units are
much easier to match to the many small pieces of electrical demand.
Even a multimegawatt wind turbine can be built so quickly that the U.S.
will probably have a hundred billion watts of them installed before it
gets its fi rst one billion watts of new nuclear capacity, if any.
Small, quickly built units also have far lower financial risks than
big, slow ones. This gain in financial economics is the tip of a very
large iceberg: micropower’s more than 200 different kinds of hidden fi
nancial and technical benefits can make it about ten times more
valuable (www.smallisprofitable.org) than implied by current prices or by the cost comparisons above. Most of the same benefits apply to negawatts as well.
Despite their small individual size, micropower generators and
electrical savings are already adding up to huge totals. Indeed, over
decades, negawatts and micropower can shoulder the entire burden of
powering the economy.
The Electric Power Research Institute (EPRI), the utilities’
think-tank, has calculated the U.S. negawatt potential (cheaper than
just running an existing nuclear plant and delivering its output) to be
two to three times nuclear power’s 19 percent share of the U.S.
electricity market; RMI’s more detailed analysis found even more.
Cogeneration in factories can make as much U.S. electricity as nuclear
does, plus more in buildings, which use 69 percent of U.S. electricity.
Windpower at acceptable U.S. sites can cost-effectively produce at
least twice the nation’s total electricity use, and other renewables
can make even more without significant land-use, variability, or other
constraints. Thus just cogeneration, windpower, and efficient use—all
profitable—can displace nuclear’s current U.S. output roughly 14 times
Nuclear power, with its decade-long project cycles, difficult
siting, and (above all) unattractiveness to private capital, simply
cannot compete. In 2006, for example, it added less global capacity
than photovoltaics did, or a tenth as much as windpower added, or 30–41
times less than micropower added. Renewables other than big hydro dams
won $56 billion of private risk capital; nuclear, as usual, got zero.
China’s distributed renewable capacity reached seven times its nuclear
capacity and grew seven times faster. And in 2007, China, Spain, and
the U.S. each added more windpower capacity than the world added
nuclear capacity. The nuclear industry does trumpet its growth, yet
micropower is bigger and growing 18 times faster.
President Bush rightly
identifies the spread of nuclear weapons as the gravest threat to
America. Yet that proliferation is largely driven and greatly
facilitated by nuclear power‘s flow of materials, equipment, skills,
and knowledge, all hidden behind its innocent-looking civilian
disguise. (Reprocessing nuclear fuel, which the President hopes to
revive, greatly complicates waste management, increases cost, and
boosts proliferation.) Yet acknowledging nuclear power’s market failure
and moving on to secure, least-cost energy options for global
development would unmask and penalize proliferators by making bomb
ingredients harder to get, more conspicuous to try to get, and
politically costlier to be caught trying to get. This would make
proliferation far more diffi cult, and easier to detect in time by
focusing scarce intelligence resources on needles, not haystacks.
Nuclear power has other unique challenges too, such as long-lived
radioactive wastes, potential for catastrophic accidents, and
vulnerability to terrorist attacks. But in a market economy, the
technology couldn’t proceed even if it lacked those issues, so we
needn’t consider them here.
So why do otherwise well-informed
people still consider nuclear power a key element of a sound climate
strategy? Not because that belief can withstand analytic scrutiny.
Rather, it seems, because of a superficially attractive story, an
immensely powerful and effective lobby, a new generation who forgot or
never knew why nuclear power failed previously (almost nothing has
changed), sympathetic leaders of nearly all main governments, deeply
rooted habits and rules that favor giant power plants over distributed
solutions and enlarged supply over efficient use, the market winners’
absence from many official databases (which often count only big plants
owned by utilities), and lazy reporting by an unduly credulous press.
Isn’t it time we forgot about nuclear power? Informed capitalists
have. Politicians and pundits should too. After more than half a
century of devoted effort and a half-trillion dollars of public
subsidies, nuclear power still can’t make its way in the market. If we
accept that unequivocal verdict, we can at last get on with the best
buys first: proven and ample ways to save more carbon per dollar,
faster, more surely, more securely, and with wider consensus. As often
before, the biggest key to a sound climate and security strategy is to
take market economics seriously.
Mr. Lovins, a physicist, is cofounder, Chairman, and Chief
Scientist of Rocky Mountain Institute, where Mr. Sheikh is a Research
Analyst and Dr. Markevich is a Vice President. Mr. Lovins has consulted
for scores of electric utilities, many of them nuclear operators. The
authors are grateful to their colleague Dr. Joel Swisher PE for
insightful comments and to many cited and uncited sources for research
help. A technical paper preprinted for the September 2008 Ambio (Royal
Swedish Academy of Sciences) supports this summary with full details
and documentation (www.rmi.org/sitepages/ pid257.php#E08-01).
RMI’s annual compilation of global micropower data from industrial and
governmental sources has been updated through 2006, and in many cases
through 2007, at www.rmi. org/sitepages/pid256.php#E05-04.
This is conservatively used as the basis for all comparisons in this
article. The ~2-3¢/kWh nuclear "production costs" often quoted are the
bare operating costs of old plants, excluding their construction and
delivery costs (which are higher today), and under cheap old fuel
contracts that are expected to rise by several-fold when most of them
expire around 2012.
- All monetary values in this article are in 2007 U.S. dollars.
All values are approximate and representative of the respective U.S.
technologies in 2007. Capital and operating costs are levelized over
the lifespan of the capital investment.
- Distributed generators may rely on the power grid for emergency
backup power, but such backup capacity, being rarely used, doesn't
require a marginal expansion of grid capacity, as does the construction
of new centralized power plants. Indeed, in ordinary operation,
diversified distributed generators free up grid capacity for other
- Solar power is not included in Figure 1 because the delivered
cost of solar electricity varies greatly by installation type and
financing method. As shown in Figure 4, photovoltaics are currently one
of the smaller sources of renewable electricity, and solar thermal
power generation is even smaller.
- A similar credit for displaced boiler fuel can even enable this
technology to produce electricity at negative net cost. The graph
conservatively omits such credit (which is very site-specific) and
shows a typical positive selling price.
- We ignore here the modest and broadly comparable amounts of
energy needed to build any kind of electric generator, as well as
possible long-run energy use for nuclear waste management or for
extracting uranium from low-grade sources.
- Since its recovered heat displaces boiler fuel, cogeneration
displaces more carbon emissions per kilowatt-hour than a large
gas- red power plant does.
- However, at long-run gas prices below those assumed here (a
levelized 2007-$ cost of $7.72 per million BTU, equivalent to assuming
that this price escalates indefinitely by 5%/y beyond
inflation-yielding prices far above the $7-10 recently forecast by the
Chairman of Chesapeake, the leading independent U.S. gas producer) and
at today's high nuclear costs, the combined-cycle plants may save more
carbon per dollar than nuclear plants do. This may also be true even at
the prices assumed here, if one properly counts combined-cycle plants
ability to load-follow, thus complementing and enabling cleaner,
cheaper variable renewable resources like windpower. Natural gas could
become scarce and costly only if its own efficiency opportunities
continue to be largely ignored. RMI's 2004 study Winning the Oil
found, and further in-house research confirmed in detail, that the US
could save at least half its projected 2025 gas use at an average cost
roughly one-tenth of the current gas price. Two-thirds of the potential
savings come from efficient use of electricity and would be more than
paid for by the capacity value of reducing electric loads.
- Data for decentralized gas turbines and diesel generators exclude generators of less than 1 megawatt capacity.
Correction: April 28, 2008
Due to new data, footnote 1 and 8 have been edited to reflect this new information.
Fair Use Notice.