The main risk is commercial capture, not whether Cerence can build useful automotive AI. The company already has product, customers, and cash generation; the harder question is whether it can hold enough control of the in-car assistant economics before OEMs, cockpit-stack vendors, or
hyperscalers compress the layer into a thinner
royalty stream. Validation timing, customer concentration, legal cost, and auto-cycle noise can all delay the
rerating.