Why is nuclear power so expensive?

Nuclear power is expensive to build, not to run. Almost the entire cost is front-loaded: a decade of construction and billions in financing, while fuel costs are close to negligible. In 2021 nuclear LCOE ran USD 131–204/MWh against USD 24–41/MWh for utility-scale solar (Lazard LCOE Analysis v16). Georgia's Vogtle plant was completed in 2024, seven years late and more than USD 20 billion over budget.

Nuclear's cost structure is the inverse of a fossil plant's. Fuel is roughly 70% of the generating cost of a gas-fired plant, but almost negligible for a reactor: once built, a nuclear plant produces very cheap electricity. The difficulty is everything that comes before that. A ten-year build and billions of dollars of financing are a feat few private firms can carry alone, which is why nuclear has never been a genuine free-market technology. The US industry exists partly because the Price-Anderson Act caps operator accident liability, and modern projects typically need government guarantees on multi-billion-dollar loans.

Whether nuclear is 'expensive' also depends on which yardstick is used. Levelised cost of electricity (LCOE) divides lifetime cost by lifetime output, and on that measure nuclear looks far dearer than recent solar or wind. But LCOE ignores system costs: solar generates only when the sun shines, so grid stability requires backup storage or peaking plant, while nuclear runs around the clock and avoids those integration costs. Levelised avoided cost of electricity (LACE) measures what the grid saves by having reliable baseload. Once renewable penetration passes 30%, the cost of balancing intermittency rises exponentially, and the case for nuclear as a grid anchor strengthens.

The third driver is the loss of build capability itself. For three decades the United States essentially stopped ordering reactors. South Carolina's V.C. Summer project was abandoned in 2017 after USD 9 billion and ten years of construction with almost nothing to show; Georgia's Vogtle plant finally reached completion in 2024, seven years late and over USD 20 billion over budget. The ageing fleet is costly too: most US reactors date from the 1970s, spare parts are long out of production, and every repair is a bespoke fabrication job, pushing operating costs above cheap natural gas.

Counter-examples show that cost overruns are not destiny. France's Messmer Plan built 56 reactors between 1974 and 1994, decarbonising an entire power sector; French electricity remains among Europe's cheapest and cleanest, proving that fleet-scale construction of a standardised design is the route that makes nuclear economics work. The UAE's Barakah plant, delivered by a Korean consortium, came in broadly on time and on budget and now supplies about 25% of the country's electricity. Nuclear also thrives in state-directed economies such as China and Russia, where returns are legally secured or the state funds projects directly, sustaining the 60-year planning horizon nuclear requires.

Nuclear has never been a true free-market technology; it is a partnership between the state and the industry — it invests in the next century, not the next quarter.

This question is covered in depth in The Full Spectrum Every Energy Source Explained — A Singapore Perspective，第六章 6.5、6.7