We, at 2600hz, think that voice cost structures are going the way of the dodo, thanks largely to IP and the migration away from copper. Voice as a service has to evolve to handle the demands of a connected world.
In my personal opinion, I see voice as the method of last resort, but the critical lynchpin of communication. SMS and email are fantastic when you can handle an asynchronous response, but when push comes to shove you better believe I'm ringing the phone.
Lastly, on the subject of WebRTC; I believe it will rock for small conferences but it's anything BUT agile at scale.
Consider a 10 person video call. If I use a polycom bridge there are 2 streams coming from any given participant; one in, one out. This is not the case with WebRTC.
With WebRTC, I have one stream to each participant; ergo 90 total stream and (at 300kbps streaming) that's 2.7mbps per participant. The numbers only get worse as you scale.
My contention is that WebRTC is a product for a mesh world and we live in a Hub n' Spoke reality. It will be hard for WebRTC to compete at scale.