Things might start to get shaky at 50, but I seriously doubt it's impossible.
As for how it feels to the CEO I don't think you're right. The company I had experience with was run as a triumvirate with one slightly more powerful member. All three knew everything that was going on and had a hand in it, but two of them were experts in different areas and one was a generalist and overall vision-setter.
The CEO spent probably 80% of his time working on real projects with individuals and teams and 20% coordinating with his junior triumvirs.
He certainly was not was reduced to sitting in his office funneling orders through two people all day.
The same company that was wonderful to work for at 40 people imploded at 150.
http://en.wikipedia.org/wiki/Dunbars_number
I think people misinterpret the concept of Dunbar's number. It isn't how many people you can keep track of -- it's how large of a team you can hold together, provided you spend massive amounts of your time on team-building. From the article:
>Dunbar has argued that 150 would be the mean group size only for communities with a very high incentive to remain together. For a group of this size to remain cohesive, Dunbar speculated that as much as 42% of the group's time would have to be devoted to social grooming.
Spending 42% of your time on social grooming isn't a very efficient way to run a company, though, especially when you only work 8 of your 16 waking hours (and that's generally considered a good thing).
No.
The reason 10 people can work like magic is that they all report to you. There are no middle managers, just people who get things done, plus one person to coordinate. Everybody knows what everybody else is doing, and the communication path is only one level.
The upper bound for this type of company is how many people you can manage personally. You can not manage 30 people, and you probably can't manage 20. Once you need to add a layer of managers under you who aren't actually getting shit done, the entire dynamic changes.