I kind of agree. Although I think the underlying assumption is that it's an infrastructure as a service play, and so Twilio et al are creating value. Obviously their target market is the web app developer crowd.

To do it yourself you need to line up your outbound SIP providers, inbound DID providers (not one and the same once you dig into the details), then you need to operate a bunch of Freeswitch / Asterisk / whatever SIP servers, and learn how to program them, and learn how keep them running smooth (Not particularly difficult, but in terms of learning requirements - it's akin to meeting a database + web server for the first time).

Due to the real-time nature of telcos (and I mean that in the traditional clock-synch'd sense of the word) there are additional complications if you want to use a virtualised OS.