SWITCHED ON
The daily technology series nobody asked for but everyone needed
Nuclear: The Comeback Nobody Expected
The energy source that was supposed to be dead is quietly becoming the hottest topic in the global power sector. Here is why, and whether the enthusiasm is warranted this time.
Nuclear power has two problems. The first is that it occasionally goes catastrophically wrong in ways that are visible, terrifying, and impossible to ignore. The second is that it mostly goes quietly right in ways that are invisible, reliable, and easy to take for granted. The public conversation has spent fifty years focused almost entirely on the first problem and almost none on the second.
Season one of Switched On ended with a meditation on what twenty-five episodes of daily technology coverage adds up to, and what the pattern running through all of it — deploy fast, govern late, pay the consequences slowly — means for how we should approach the next wave of consequential technologies. Season two begins with one of those technologies, and it is one with a longer history than almost anything else we covered. Nuclear energy. The source of power that was supposed to define the twentieth century, nearly destroyed itself twice in living memory, was written off by most of the developed world, and is now — quietly, stubbornly, and with more serious backing than at any point in the past three decades — staging a comeback that deserves considerably more attention than it is getting. Welcome back. We have missed you. Let's get into it.
01 — What Nuclear Actually Is and Why It Was Different
Nuclear power generates electricity through fission — splitting heavy atoms, typically uranium-235 or plutonium-239, releasing energy as heat, which drives steam turbines in a process that is, at that mechanical level, identical to a coal plant. The difference is the scale of energy density. A kilogram of uranium fuel contains roughly three million times more energy than a kilogram of coal. This extraordinary energy density is what makes nuclear power uniquely capable of generating large amounts of continuous, weather-independent, low-carbon electricity from a very small physical and resource footprint. A nuclear plant generating one gigawatt of electricity — enough to power roughly 700,000 homes — occupies a few square kilometres. The equivalent solar capacity would require roughly 75 square kilometres of panels. The equivalent wind capacity would require significantly more.
Nuclear's post-war promise was extravagant. Electricity "too cheap to meter" was the phrase — a slogan that did not survive contact with construction cost reality but captured the genuine excitement of a technology that seemed to offer limitless energy from a fuel source that weighed almost nothing. By the 1970s, nuclear was providing a growing share of electricity in the US, France, Japan, and other developed economies. Then came Three Mile Island in 1979, Chernobyl in 1986, and the combination of cost overruns, regulatory tightening, and public opposition that made new nuclear construction in most Western countries economically and politically untenable for a generation.
Fukushima Daiichi in 2011 — a meltdown triggered by the Tōhoku earthquake and tsunami — accelerated the retreat. Germany announced an immediate nuclear phase-out. Japan shut down its entire fleet pending safety reviews. The narrative of nuclear as a dying technology, appropriate perhaps as a legacy asset but not as a future investment, seemed settled.
Nuclear power has killed fewer people per unit of energy generated than almost any other source in history, including solar and wind when full lifecycle manufacturing and installation deaths are counted. This fact has had almost no effect on public perception, which is a case study in how risk communication works and why it so often fails.
02 — Why the Narrative Is Shifting
Three forces are driving nuclear's rehabilitation simultaneously, and their convergence is what makes this moment different from previous periods of nuclear enthusiasm.
The first is the climate imperative. Decarbonising electricity grids by 2050 — the stated ambition of most developed economies — requires enormous volumes of low-carbon, dispatchable power. Renewable energy's intermittency problem, as we covered in Season One's climate episode, means that a grid running primarily on solar and wind requires either massive storage capacity, long-distance transmission infrastructure, or dispatchable backup generation that can run when the sun does not shine and the wind does not blow. Nuclear is the only low-carbon source that provides this dispatchability at scale. The IPCC's scenarios for limiting warming to 1.5 degrees virtually all include a significant role for nuclear. Climate scientists who oppose nuclear on principle are a minority of the field. The majority view among energy researchers is that ruling out nuclear makes the net-zero target harder and more expensive.
The second force is the AI electricity demand explosion. Data centres running AI workloads are extraordinarily power-hungry and growing fast. Microsoft, Google, Amazon, and Meta have all signed agreements with nuclear operators and developers in the past two years, seeking long-term, reliable, low-carbon electricity supplies for their expanding computing infrastructure. Microsoft's deal to restart the Three Mile Island Unit 1 reactor — the unit that was not involved in the 1979 accident and had been shutting down for economic reasons — is the most symbolically striking of these. The technology industry, which has significant political and economic influence and a net-zero commitment, has become one of nuclear's most important new constituencies.
The third force is the emergence of small modular reactors — SMRs — as an alternative to the enormous, bespoke, extraordinarily expensive gigawatt-scale plants that drove nuclear's cost crisis. We will return to SMRs in detail shortly, because they deserve their own section.
03 — The Cost Problem and Whether SMRs Solve It
Nuclear's fundamental economic problem in the past three decades has been cost. Large nuclear plants have been plagued by construction overruns of staggering magnitude. The Vogtle expansion in Georgia — the first new nuclear reactors built in the US in decades — came in at roughly $35 billion for two reactors, approximately twice the original budget, and seven years late. Hinkley Point C in the UK is tracking toward £46 billion, having started at £18 billion. These are not isolated failures. They are the consistent pattern of large nuclear construction in Western countries, driven by one-off bespoke designs, stringent regulatory requirements applied inconsistently, supply chain atrophy from decades of no new construction, and the absence of the learning-curve effects that come from building the same design repeatedly.
Small modular reactors propose a different model. Reactors of 50 to 300 megawatts — small by conventional standards — designed for factory manufacture and on-site assembly rather than bespoke field construction. The theory is compelling: factory production enables quality control and learning-curve cost reductions similar to those that drove down solar panel costs; standardised designs reduce regulatory uncertainty; smaller size reduces financial risk per project; modular scalability allows capacity to be added incrementally rather than in enormous single bets.
The reality is that SMRs remain, as of 2026, largely unproven at commercial scale in Western markets. NuScale, the US company that was furthest along with a certified SMR design, cancelled its first commercial project in 2023 after cost estimates rose to levels that made the economics unworkable against natural gas and renewables. Rolls-Royce in the UK, GE-Hitachi with its BWRX-300, and several others are progressing through regulatory approval processes but have not yet demonstrated the cost reductions the model requires. China and Russia have operating SMRs, which provides proof of concept for the technology without proving the Western regulatory and commercial case.
The honest assessment: SMRs may deliver on their promise, but the learning curve requires building many of them first, and the investment required to get to the point where the cost reductions materialise is substantial. This is not an argument against proceeding. It is an argument for proceeding with accurate expectations rather than the optimistic projections that have characterised nuclear advocacy before and occasionally damaged its credibility.
04 — The Waste Problem, Honestly
Nuclear waste is the argument that never goes away, and it deserves honest treatment rather than the dismissal it sometimes receives from nuclear advocates or the catastrophising it receives from opponents.
The volume of nuclear waste is, by the standards of other industrial processes, remarkably small. All the high-level nuclear waste generated by the entire US nuclear industry over six decades would fit within a single Walmart store. This is a genuine and important fact. It reflects nuclear's extraordinary energy density — a very small amount of fuel generates a very large amount of electricity and leaves a very small volume of waste.
The challenge is not the volume. It is the duration and the toxicity. High-level nuclear waste remains hazardous for tens of thousands of years. No country in the world has yet opened a permanent deep geological repository for high-level nuclear waste, though Finland is furthest along with its Onkalo facility, which began construction and is the most advanced permanent repository in the world. The US nuclear waste programme has been effectively paralysed for decades by the political failure of the Yucca Mountain repository project in Nevada, leaving spent fuel stored in temporary dry cask storage at reactor sites across the country — a situation that is safe in the short term and inadequate as a long-term solution.
Advanced reactor designs, including some SMR concepts and fast reactor designs, can use existing spent fuel as fuel themselves, significantly reducing both the volume and the longevity of the waste produced. This is not yet commercially operational at scale anywhere in the world. It is a genuine future possibility that nuclear advocates are right to point to and wrong to treat as a solved problem.
05 — Where This Is Actually Going
The global nuclear fleet is, for the first time in decades, growing rather than shrinking. China is building more nuclear reactors than the rest of the world combined. India has an ambitious expansion programme. South Korea has reversed its phase-out policy. Japan is restarting reactors and extending operating licences. France, which generates approximately 70 percent of its electricity from nuclear and briefly considered reducing that share, has recommitted to the technology and announced new build. The UK government has designated nuclear as critical infrastructure and is supporting new projects. The US Inflation Reduction Act included production tax credits for existing nuclear plants that have kept economically marginal reactors open.
The trajectory is clear. Whether it is fast enough to matter for near-term climate targets is a harder question. New nuclear takes a long time to build — even optimistic SMR timelines put first commercial units in the early 2030s for Western markets. The existing fleet, kept running through life extensions, is probably the most important near-term contribution nuclear can make to decarbonisation. The new build pipeline matters most for the 2040s and beyond.
What has definitively changed is the political and intellectual landscape. Nuclear scepticism, once the default position of the environmental movement, is no longer consensus. A growing number of prominent environmentalists, climate scientists, and energy researchers have shifted toward cautious support for nuclear as a necessary component of the clean energy transition. The debate is no longer between nuclear advocates and a united opposition. It is a genuine argument about costs, timelines, and priorities among people who share the goal of decarbonisation. That is a more productive argument than the one that preceded it, and it is happening now.
Tomorrow we are going somewhere that sits at the intersection of nuclear's comeback and the broader question of who controls the world's most consequential technologies — semiconductor geopolitics. The chips that run everything: where they are made, why that matters more than almost anything else in the current geopolitical landscape, and what happens if the supply chain breaks. See you then.
Switched On is a daily technology series covering the ideas, systems, and arguments shaping the digital world. Opinionated. Witty. Occasionally wrong. Always worth the argument.
