The trouble with small nuclear reactors

Small modular reactors (SMRs) are being promoted as a reliable, low-carbon source of electricity for grids facing rising demand, including from AI data centres. Governments and investors are backing projects such as Ontario’s Darlington plant, where four reactors are planned at an estimated cost of $15bn. Advocates say smaller units can be built faster, occupy less space and become cheaper through factory production and repeated construction. But those economies have not been demonstrated, and first-of-a-kind projects carry serious cost and schedule risks. Developers’ estimates have risen sharply, while NuScale’s proposed US project was cancelled after costs climbed and buyers balked at committing to uncertain prices. Critics argue that public money and ratepayer funds could instead support renewables, batteries or established nuclear designs.

The economics remain uncertain: one recent study found SMRs most likely to compete for industrial customers and with US tax support, while some developers’ projected power costs exceed those of solar and may rival or surpass large nuclear plants. Supporters counter that constant, flexible power has value beyond simple price comparisons, especially for data centres willing to pay more. Yet whether reactors will operate reliably and deliver promised savings is still unproven. Fuel may be an equally serious constraint: several designs need highly enriched uranium, or HALEU, with little commercial supply outside China and Russia. Governments are funding domestic enrichment, but supplies are limited, and producing HALEU requires more uranium. The technology’s prospects therefore depend not just on investment, but on proving its costs, performance and fuel supply.