A new study in Nature Energy estimates fusion power's experience rate at 2-8 percent, indicating slower cost declines than assumed in most energy models (Tang et al., Nature Energy, 2026; MIT Technology Review, April 23, 2026).

Researchers at ETH Zurich interviewed fusion experts from public and private sectors on unit size, design complexity and customization requirements for magnetic confinement and laser inertial confinement approaches. Experts rated complexity as extremely high and plants as large, similar to coal and fission facilities, yielding the 2-8 percent range. This exceeds fission's documented 2 percent experience rate but trails solar modules at 23 percent, lithium-ion batteries at 20 percent and onshore wind at 12 percent (Tang et al., Nature Energy, 2026; MIT Technology Review, April 23, 2026).

Current modeling studies frequently assume 8-20 percent experience rates for fusion, according to the paper. U.S. public funding exceeded $1 billion in fiscal year 2024 while private-sector investment totaled $2.2 billion from July 2024 to July 2025 (MIT Technology Review, April 23, 2026). Separate IEA data show data-center electricity consumption, accelerated by AI training, is projected to double by 2026; Goldman Sachs analysis estimates AI-related demand could reach 10 percent of U.S. power generation by 2030 (IEA Electricity 2024; Goldman Sachs, 2024).