I agree as someone who's wanted to develop an iOS app for years but can't get over the App Store monopoly risk.
But, from a business sense, doesn't it seem like they're doing it backwards? Why not make a best-in-class AI assistant app, build a big userbase, and take the money to research hardware and harass Apple concurrently? Why start with hardware first?
An example: Google/Apple doesn't allow third-party apps access to call audio. So you couldn't implement live translation/transcription, by just being an app.
Google/Apple will jealously guard their turf in their ecosystems (to make space for their own AI assistants probably), so this level of third-party access is unlikely to be forthcoming.