I've long thought that the end-game for voice control is an omnipresent interaction. For example, using Siri across my watch, phone, computer, or speaker seamlessly and with shared context. Whether that is achieved through device-to-device communication (most likely Apple's approach) or device-cloud-device interaction is pretty irrelevant to the end game of voice control.

I actually agree with Phil Schiller's comment that most of the use cases for voice control aren't facilitated by a device without a screen. [1] Apple's advantage is that the design of a HomePod can basically assume that a nearby iPhone or iPad is available without having to launch an app.

Somewhat ironically, Nest actually seems to have one of the best approaches I've seen to architecting a platform for such omnipresent interactions to take place on, with Nest OpenWeave[2].

[1] - https://gadgets.ndtv.com/apps/features/phil-schiller-upgrade... [2] - https://github.com/openweave/openweave-core