When you say, “large” companies, exactly how large are we talking about?
For the most part, Asterisk installations tend to be what I consider small. (less than 50 or so lines)
A good illustration of this is the success of Asterisk@Home. (Which is arguably the most used implementation.) All of it’s implementations are SOHO.
Having said that, most every VOIP provider on the internet has some version of Asterisk under the hood, so even “service provider scalability” is possible. However, each had to do major league alterations to the underlying code, none of which they put back into the project.
If you’re looking to put this in a company that has 500 or so employees, you’re in for HUGE project. Yeah, it’s less expensive, but that’s because you’re doing all the work. You design, build, configure, re-write code, and support the whole system from start to finish with little or no documentation, scant support (from digium), and no opportunity for any sort of training. If you don’t watch out, putting an Asterisk system in service, and keeping it online could easily become all that you do.
Present it as an option, but be realistic about it’s capabilities, reliablity, and your ability to keep it running with the same uptime that your management has become used to with other voice systems.
Ultimately, every system you buy involves risk and expense of some sort. Your management has to decide for itself if the cost savings justifies the risk. If mitigating the risk involves buying a system from someone who has promised 24 hour repairs with a 4 hour response time to problems, then so be it.
You’ll probably find that the decision is less about how many other companies have decided to drink the kool-aid (so to speak), and more about your management’s confidence in your own ability to pull it off.