Platforms have two main stakeholders (besides the platform maker) — users and partners. In the case of physical devices, more than being about users, it’s about endpoints. From how many physical devices can users access the service? It’s also about how many different services have been integrated through its API.
For Alexa, the endpoints are Echos (vanilla, Dot, Tap, Look, Show) and Alexa Voice Service products (the likes of Nucleus, Triby, Jam, new HTC phones, Fire TVs/tablets, etc). For Google Assistant, it’s almost any Android phone/device, the Google Home, and soon Google Assistant SDK-enabled hardware.
In terms of the APIs, it’s Actions on Google vs Alexa Skills Kit. For any other service that might come online as an embeddable entity, maybe Cortana or Siri, it’s likely to follow the same pattern. The value of the network to developers increases with the number of users and endpoints. With increased competition of offering, we’re likely to see more incentives offered to hardware makers to integrate Alexa, Google Assistant, or other services.