The Nuclear Reactor Trap: Sensitive to Droughts but Essential for AI
The surge in energy demand driven by AI is reviving interest in nuclear power, yet global warming and falling river levels are undermining the reliability of nuclear plants, forcing output cuts.
The growing demand for energy and the rapid acceleration of artificial intelligence are leading the world back into a nuclear energy arms race. However, global warming poses a significant obstacle, primarily due to the drop in river levels required to cool nuclear power plants, which threatens energy stability.
As Europe faces extreme heat and peak demand for air conditioning, nuclear reactors at approximately six sites have been shut down or reduced output in recent months. France, which recorded a 20% drop in output, has seen a series of reactors suspend operations. Similar issues have affected Central and Eastern Europe, where the drought lowered the Danube to historic lows, impacting the Paks plant in Hungary, the Cernavodă reactor in Romania, and the Beznau facility in Switzerland.
The Energy Dilemma
Nuclear reactors require massive volumes of water for cooling, which is then discharged back into the environment. During heatwaves, river water temperatures rise, making it impossible to discharge the water without harming the ecosystem. Three heatwaves this year have already forced production cuts, mirroring the 2022 crisis that caused electricity prices to skyrocket in France, Italy, and the UK.
Despite these challenges, many European nations are shifting policy. Italy is preparing to lift its long-standing ban on nuclear construction, Belgium and Sweden are reversing previous phase-out decisions, and the UK is streamlining regulations to promote new projects. While many hope nuclear power will provide the carbon-free energy needed for AI data centers, experts warn of long-term reliability and economic risks.
Timelines and Costs
Building a nuclear power plant takes an average of 14.5 years, whereas wind and solar farms can be built in five years or less. According to Lazard, the Levelized Cost of Energy (LCOE) for new nuclear is $155 per MWh, compared to $43 for onshore wind and $41 for commercial solar.
Furthermore, the industry faces the monumental challenge of storing nuclear waste for hundreds of thousands of years. Even modern reactors remain susceptible to cooling system failures, with past disasters like Chernobyl and Fukushima highlighting the risks of radioactive contamination and the need for permanent area evacuations.
The Promise of Small Modular Reactors (SMRs)
Small Modular Reactors (SMRs) are being touted as a more economical solution for AI data centers. However, the technology remains unproven commercially. The cancellation of the NuScale project in Idaho in 2023, following a price surge from $58 to $89 per MWh, served as a major setback for the industry. For now, renewable energy remains a more immediate and cost-effective solution to the world's growing energy insecurity.
