Nuclear's Real Problem is the Spreadsheet, not Safety.


Nuclear power is enjoying a revival. Governments are reversing phase-outs, and technology companies hungry for round-the-clock clean electricity are signing deals to restart old reactors and fund new ones. Microsoft agreed to buy power from a restarted unit at Three Mile Island, while Google and Amazon have backed developers of small modular reactors. Yet the biggest obstacle to a nuclear renaissance has little to do with meltdowns or waste. It is cost. In the West, new nuclear plants have become some of the most expensive infrastructure projects on Earth, and understanding why is essential to judging whether the revival can last.
The recent record is sobering. Georgia's Vogtle Units 3 and 4, the first new reactors completed in the United States in decades, came online years behind schedule at a total cost of more than $30 billion, more than double the original estimate. In South Carolina, the V.C. Summer project was abandoned in 2017 after roughly $9 billion had been spent, leaving ratepayers paying for reactors that will never produce a watt. Britain's Hinkley Point C has followed a similar arc of delays and escalating budgets. In 2023, NuScale cancelled what was meant to be America's first small modular reactor plant after projected power prices rose and too few utilities signed on.
The core problem is financial structure. A nuclear plant is almost entirely an upfront capital cost; fuel is cheap, but the plant must be built, and paid for, long before it earns anything. That makes nuclear acutely sensitive to interest rates and delays. Every year of construction adds financing costs on billions of dollars that are not yet generating revenue, so a schedule slip can inflate the final bill far more than the extra labor and materials alone. As earlier posts on this blog have argued, the discount rate quietly decides many climate questions; for nuclear, it may decide everything. Compare this with solar and wind, which are built in months, in small modular increments, and begin earning almost immediately.
Nuclear also suffers from what economists call a negative learning curve. Most technologies get cheaper as more units are built, which is exactly what happened with solar panels and batteries. Nuclear in the West has often done the opposite: each project is a near-first-of-a-kind effort, with designs changed midstream, supply chains rebuilt from scratch, and specialized workers who have not built a reactor in a generation. The exceptions are revealing. South Korea kept costs relatively stable for decades by building the same standardized design repeatedly with the same teams, and its exported reactors at Barakah in the United Arab Emirates were completed far closer to budget than their Western counterparts. The lesson is that nuclear's costs are not fixed by physics; they reflect how a country builds.
Whether the current revival succeeds depends less on new reactor designs than on fixing that industrial model. Small modular reactors promise factory production and repetition, but they must be ordered in large numbers to capture those savings, and that requires patient buyers willing to absorb early overruns. Tech companies with deep pockets and long horizons may be exactly those buyers, and public loan guarantees can lower financing costs. If governments and investors commit to standardized designs built in fleets rather than one-off megaprojects, nuclear could become a reliable complement to renewables. If not, it risks repeating its history: a clean technology that keeps losing to its own balance sheet.



Comments