.@scottgmorton spent nine years writing control software at SpaceX. He left to build the software he wished he'd had. That instinct cuts against the dominant narrative that AI can solve software. So where does that leave companies building for nuclear reactors, rocket engines, refineries? Morton agrees shallow web apps are likely in trouble. But high-consequence infrastructure, he says, is different: "Nobody is going to bet their career on vibe-coding a nuclear reactor or a rocket engine,…