Restarting Retired Nuclear Plants Costs $1.5-1.8 Billion Each and Has Never Been Done Before in the U.S., Creating Unprecedented Regulatory and Engineering Risk

The U.S. is attempting to restart retired nuclear reactors for the first time in history — Constellation's Three Mile Island Unit 1 (Crane Clean Energy Center) at an estimated $1.6 billion, and Holtec's Palisades plant in Michigan at $1.5 billion in DOE loans plus $300 million from Michigan — but no country has ever returned a fully decommissioned commercial reactor to service, and the NRC has no established regulatory process for reactor restarts. Why it matters: these projects are being driven by AI data center electricity demand (Microsoft signed a 20-year PPA for TMI-1's output), so if the first restart projects fail or dramatically overrun, it will discredit the entire restart pathway for the 12+ additional retired reactors being evaluated, so the data center industry's nuclear ambitions will collapse back to natural gas, so the political narrative that nuclear can be deployed quickly (restarts) rather than slowly (new builds) will be undermined, so the bipartisan nuclear enthusiasm of 2024-2025 could evaporate before translating into actual megawatts. The structural root cause is that reactor restarts require re-qualifying aged components that have been exposed to years of non-operational conditions (thermal cycling, corrosion), rebuilding institutional knowledge after operating staff dispersed, and navigating an NRC licensing framework that was designed for continuous operation or permanent shutdown — not for resurrection.