Why big tech's nuclear plans could blow up (www.bbc.com)

Big tech is increasingly investing in nuclear power—both buying existing reactor output (Microsoft’s deal for power from Three Mile Island) and funding next‑gen small modular reactors (SMRs)—to secure the “always‑on,” carbon‑free electricity AI workloads demand. Firms such as Google and Microsoft view nuclear as a strategic lever to supply huge data centers and high‑power AI chips as electricity needs surge: data centers already consume ~1.5% of global power and, the IEA warns, that could double within five years. Tech money and lobbying are accelerating pilots and partnerships (e.g., Google with Kairos Power aiming for ~50 MW by 2030), but the industry is betting on a technology that is largely still experimental. The catch is practical: SMRs trade economies of scale for modularity, which tends to reduce thermal efficiency and raise per‑megawatt costs and—even per a Stanford study—can generate more radioactive waste than larger reactors. Commercial rollout, licensing and public acceptance (Three Mile Island’s legacy, local bans like North Tonawanda) mean SMRs are unlikely to close the immediate AI energy gap. That leaves a mixed path forward: long‑term nuclear may be part of a decarbonized grid, but near‑term relief will depend on tech fixes—more energy‑efficient chips, leaner models, better demand management—and hard economic choices about who pays for scarce, growing power.