Taiwan Apopyrenistes
Taiwan is struggling to denuclearize, but the real error occured decades ago
Taiwan recently announced that it was going to decommission its last nuclear plant. When I first heard of the decision, my initial reaction was horror. I wrote that it made me begin to doubt whether America should defend an island that appears unwilling to try to protect itself from Chinese energy blockades.
In one sense, I was wrong. Why? Keeping the current reactors simply doesn’t matter that much. Nuclear power is a sideshow, providing only 4% of electricity demand in 2024.
But in a deeper sense, Taiwan’s reluctance to nuclearize decades ago shows a lack of seriousness about security issues. The country should have gone nuclear decades ago. It would have generated economic benefits — in other words, it would have been costless — and provide massive security externalities.
But it did not happen. And that is the problem, much more than the government’s decision to shut down a small marginal nuclear power sector which barely registers in the face of the island’s skyrocketing power demands.
Still, the story isn’t a simple morality play. Taiwan’s nuclear program was born under conditions that would have made it very difficult to sustain once the country became a democracy, regardless of the economic or strategic merits.
Shēng lái yǒu zuì
Taiwan’s nuclear program began as a quest for the Bomb. U.S. intelligence first got worried in 1965, when Taiwanese officials made a suspicious visit to Israel. In 1966, American analysts reported that Taiwan hadn’t pulled the trigger on a program but had decided to build atomic power plants in order to have the capacity to roll out atomic weapons if necessary. The same report, however, noted that Taipei’s plan was much more ambitious than what you’d need to have the capability to build a fission bomb on demand: the first commercial reactor would be up and running by 1972, followed by one more reactor every two years … “for an indefinite period.”
In addition, the Taiwanese planned to build a reprocessing plant. Reprocessing plants take spent nuclear fuel rods, reconcentrate the fissile material, and send it back to the power plant to use again. You can absolutely see why that would be attractive for a small island vulnerable to blockade. Except, well, reprocessing plants are also useful for concentrating fissile materials to make atomic bombs.
The United States wasn’t buying it. In 1973, the U.S. Embassy in Taipei bluntly told Washington that the Taiwanese were trying to conceal their pursuit of atomic weapons “behind the face of a plan ostensibly designed to serve Taipower’s mid-eighties fuel requirements.” The Carter administration threatened to abandon the island and the Taiwanese backed down. (Links go to primary sources at the GWU National Security Archive.)
Of course, it didn’t end there. The Taiwanese tried again after the U.S. terminated the defense treaty, but it turned out that the deputy director of their nuclear program worked for the CIA. The Reagan administration snuck him out of the country and once again threatened the Taiwanese (albeit quietly — no tariffs were involved) and this time Taiwan genuinely abandoned its quest for atomic weapons.
In the course of developing its bomb energy program, Taiwan opened up nuclear power plants in 1978, 1981, and 1984. Not quite one every two years, but close. By 1985, state-owned Taipower had already spent $78 million (NT$3.1 billion) on preparation for a fourth plant. The times they were a-changin’, however — the ruling Kuomintang still controlled the legislature, but not its own legislators. In the face of growing opposition, 55 legislators requested a temporary suspension. The executive granted it … and then Chernobyl happened. The suspension was extended to 1990.
But opposition grew rather than faded. Anti-nuclear sentiment was fueled by scandals over the secret bomb program and combined with opposition to authoritarian rule. It was a heady mix. The new opposition DPP announced its opposition to nuclear power. People protested in the streets, local governments held multiple referenda, DPP legislators went on hunger strikes and delayed bills. The lower house even passed a bill in 1996 that would end all further nuclear construction by a vote of 76 to 24 — the executive vetoed it and somehow got a new bill authorizing construction to pass 83-0. (I would love to read about how they accomplished that.) Nonetheless, construction didn’t restart until 2000, seventeen years after groundbreaking and 14 years after it was originally supposed to have opened.1
The Lungmen plant never did open. Taiwan’s opposition party, the DPP, turned anti-nuclear sentiment into practically a founding principle. It won the presidency for the first time in 2000 and stopped construction, reluctantly restarting the next year. Public protests made it difficult to bring in components. For reasons unclear, Taipower wound up in litigation with the plant’s designers. The opposition called multiple referendums, which plant proponents lost. Taipower ultimately gave up in 2021.
Nuclear capacity peaked in 1985 at 5.05 gigawatts until the government started phasing it out in 2017.
As a result, nuclear power stagnated while Taiwan’s electricity demand skyrocketed. The below chart shows Taiwan’s energy mix. Oil consumption climbed to pick up the slack until 1990 — which is absurd, because oil is a crazily expensive way to generate electricity. It made sense only because Taipower thought that nuclear construction was going to restart and didn’t want to invest in capital-intensive thermal plants. Once they realized that wasn’t going to happen anytime soon, the company switched to building coal plants. When LNG production began to take off, Taipower started building natural gas plants. At first the gas came from Indonesia and Malaysia; today it comes mostly from Qatar and Australia.
In other words, nuclear power became a sideshow long before 2017. In a country where electricity demand almost sextupled in 35 years, all you needed to do to make nuclear easy to kill was to not let it grow. The reason was that Taiwan’s nuclear program was born in under a tyrannical government that wanted to build atomic bombs. That gave it a dual stigma and made it unsurprising that the democratic opposition would oppose anything atomic. Maybe it really was Chernobyl what killed the nuclear energy program, but I’m not so sure.
The Mandarin quote that opened this section literally translates as, “original sin.”
Zǒu guò de lù, yǔ méi zǒu de lù
Thus said the LORD:
Stand by the roads and consider,
inquire about ancient paths:
Which is the road to happiness?
Travel it, and find tranquility for yourselves.
But they said, “We will not.”
—Jeremiah 6:16 (The Jewish Study Bible)
The students in my energy course covered all the above in an end-of-term paper written in December 2024.2 They estimated the cost of shutting down nuclear power for Taiwan. They came up with a solid model and did a pretty good job.
The number isn’t that large as a share of GDP. The biggest short-term issue would be the need to bring more CCGT plants on-line to provide peak capacity; you could replace nuclear baseload mostly from running existing natural gas plants more.3
They had to identify non-economic costs and benefits but they were not to address any of them — that was in the world of normative value judgments, which was above the course’s pay grade.4
But the cost of shutting down nuclear isn’t all that interesting. Taiwan can pay it. The more interesting question is how much it would have cost to go the way of France and build out nuclear power to the maximum?
The below counterfactual assumes that Taiwan continues to build at the planned rate, rolling out one 936 MW pressurized water reactor (PWR) every three years, with a six-year construction lead time. Nuclear electricity reaches 69% of demand by 2023. This is actually ahead of France by 4 percentage points, although that’s not quite fair, since French production was depressed by weather in 2022 and strikes in 2023. French output peaked at 79% in 1997; counterfactual Taiwan won’t quite match that.
This counterfactual would entail 31 reactors across 16 sites, as opposed to the six reactors across three sites that Taiwan actually constructed.
What about cost? Well, that’s harder to determine. In the real world, in most countries, the cost of building nuclear power plants escalated dramatically after the 1980s, mostly due to increasing safety concerns. (Whether that spending was worth it is another story.) Taiwanese cost data isn’t available, so we will need to make some guesses.
French reactor construction costs rose over time, whereas Korean costs fell. Roughly speaking, the cost of building a reactor in France (in 2024 dollars) rose from around $1.9 million per MW to $3.9 million per MW. In Korea, however, costs fell from $3.4 million per MW to $2.3 million.5 There are dissertations to be written about why Korean costs dropped whereas French costs rose. Luckily for this extraordinarily back-of-the-envelope calculation, it doesn’t matter much which way you do it: the counterfactual build-out would have cost Taiwan about $65 billion. In other words, nuclear wins even if we assume rising nuclear costs, a la France.
Of course, we need to subtract the costs of the coal plants that didn’t get built! Assuming roughly $800,000 per MW for a coal plant, that comes to $27 billion in coal plants that weren’t constructed.
But there’s more — coal costs money. (Shocker!) Using IMF data on the price of Australian coal, and assuming you need to burn about a half-ton to generate MWh of electricity, building out nuclear would avoid $58 billion in coal imports over the period. Of course, nuclear reactors also require fuel, but much less — roughly $700 million worth.6
In other words, going nuclear would have saved money: $65bn in construction + $0.7 billion in uranium − $27bn in coal plants − $58 billion in coal = negative $19 billion. In net present value terms, assuming a 10% discount rate and French costs, the counterfactual levelized cost of nuclear electricity (in 2024 dollars) comes to $40 per MWh, versus $50 for the coal plants actually constructed.
In short, there was a solid business case for going ahead with the nuclear program, even before you start worrying about a Chinese blockade. And once you do start worrying about a blockade, the logic becomes unassailable. Taiwan could keep the lights on indefinitely (uranium is easy to stockpile), avoid market risks, and perhaps even develop another powerful export industry. Moreover, while the Chinese could potentially target the plants in the event of a war, there are two things to keep in mind.
First, the Chinese intend to annex Taiwan, not destroy it. That makes them unlikely to deliberately try to destroy the reactors, in the same way that the Russians have not directly tried to destroy Ukrainian power plants.
Second, nuclear power plants are fairly hardened targets. It would take a direct hit to seriously damage one, as this paper shows. When the Yugoslav Air Force threatened the Krško nuclear plant during the 1991 secession crisis, the Slovenian operators reported that a cold shutdown of the facility, if carried out in time, would prevent any serious release of radioactivity even in the event of a power loss. Chinese forces would need to deliberately target the plants in a surprise attack, the way the Israelis targeted the Osirak reactor in Iraq, in order to damage them enough to threaten a serious radiation release. But that takes us back to point one.
The Taiwanese people had a path. But they said, “We will not.”
Requiem
How likely was it that Taiwan could have constructed that many nuclear plants and replicated Korea’s cost experience? At the end of the day, France is the only country to have built out atomic power to the extent envisioned here. And there is no guarantee that Taiwan would replicate the Korean cost declines. Even France saw costs explode disastrously with its latest reactor design, which finally connected in December 2024, twelve years behind schedule. Sixteen nuclear sites across the Taiwanese coast is not inconceivable, but it seems unlikely.
We’ll never know. Taiwan’s program was born in sin and never escaped the shadow of its origin. It became a political orphan, reviled by democrats, starved by bureaucrats, and ultimately abandoned by all.7
Shutting off nuclear power is a bad idea economically, but with nuclear generating only 4% of the island’s electricity, the Taiwanese will not suffer that much. Nor does the country’s small nuclear sector do much to lessen its strategic vulnerabilities. The real failure isn’t mandating the end of nuclear power. Rather, it was failing to commit to serious nuclear buildout when it could have made a difference. Taiwan needed a much bigger nuclear sector, and it’s quite possible that it could have built one for less than what it spent on coal and natural gas. That would have delivered cheap electricity. But it would have also eliminated one of Beijing’s clearest pressure points in a crisis. Right now, that would likely take the form of a “quarantine,” or an outright blockade. In the future, however, an even more powerful China could cut off Taiwan’s energy simply by telling its satraps to stop providing it. I would hope that the United States will remain strong enough to prevent that … but why should Taipei have to make a bet on American strength or willingness to stare down the People’s Republic?
A dense network of hardened reactors, fed by stockpiled uranium, would have made energy coercion a non-starter and given Taiwan the ability to ride out a long blockade. It would have demonstrated that Taiwan’s obvious capacity to engage in long-term strategic thinking applied beyond semiconductors. It built a world-beating industry in a dangerous neighborhood. It could have done the same with atomic energy.
By the time Taiwan shut its last nuclear facility in 2025, it had already forfeited the benefits. The mistake wasn’t shutting down the last visible atomic power plant. The mistake lies in all the invisible atomic power plants, the ones that we cannot see because they were never built.
The story is told in Hong Tian and Gang Lin, “Political Maneuvers on the Fourth Nuclear Power Plant in Taiwan,” China Review (November 2023) and Gillan Chi-Lun Huang, Rung-Yi Chen, and Byung-Bae Park, “Democratic innovations as a party tool: A comparative analysis of nuclear energy public participation in Taiwan and South Korea,” Energy Policy (June 2021). See also Ming-Sho Ho, “The Politics of Anti-Nuclear Protest in Taiwan: A Case of Party-Dependent Movement (1980-2000)” Modern Asian Studies (July 2003).
Did A.I. help them write it? Probably. The prose was pretty good. On the other hand, when I tried to shove their argument through Deep Research, telling it expand the basic thesis into a paper, I got gobbledygook and regurgitated Wikipedia articles.
It gets more complicated! Everything about electricity is complicated. That’s what makes it so much fun! I mean, it’s not if you’re stuck in a blackout in Spain and need to figure out just what the hell “inertia” means in the context of massless electrons but seriously, it’s fun.
They identified three issues in play: strategic defense, global warming, and emotional antinuclear sentiment.
The article is Jessica Lovering, Arthur Yip, and Ted Nordhaus, “Historical construction costs of global nuclear power reactors,” Energy Policy, Volume 91 (April 2016), pp. 371-382. You can find the data here.
This assumed a rule-of-thumb of roughly 9 MWh of electricity production for every ton of triuranium octoxide (U3O8), which the form uranium is usually traded and transported in.
After I set this thing for auto-release but before it went out, I discovered Angelica Oung’s substack, Taipology: particularly this post:
I think it supports my contention that nuclear power became a focal point for the opposition in Taiwan for reasons unrelated to the technology itself. The above post pointed in the direction of her other substack, Elemental, which is all about nuclear energy and looks excellent.