There's little doubt that in the near future all of this will move 'into the cloud'.
Microsoft has seen this and is currently rolling out its Azure cloud network. In the world of 'free' (see previous post) this, and the ability to sell advertising around everything, is the only chance the company has to maintain its predominance.
But it's not as easy as that; take the VidZapper network: to make the system as redundant as possible the most effective thing to do is to isolate functions physically on a network and to give them their own servers (you always need at least two for redundancy); when you have development, staging (or testing) and live servers this quickly becomes a lot of boxes.
You need web servers, database servers, licensing servers, player hosting servers and territorial control servers. You need firewalls and stats servers (these are process hungry so need to be separated out); you need live servers and streaming servers, possibly different ones for each format you're hosting. You also need servers to monitor your servers.
The picture's clear, it's a hugely capital expensive game which constantly needs monitoring, management and investment.
Abstracting this to 'the cloud' makes a lot of sense, but estimating the costs involved can be extremely difficult. A monthly hosting cost becomes a number of transactions. Try and figure this out.
This is important to Internet TV because the cost of delivering the content is significant compared to traditional TV. A large number of video management system also use bandwidth charges directly or indirectly as their major revenue streams (no, their services aren't actually 'free').
The first major business, with billion dollar market caps, to emerge from the rise of Internet TV was the CDN (content delivery networks) business, and this may be decimated by the rise of progressive and adaptive downloads and a move towards 'the cloud'. But the competition in the cloud marketplace is already intense and this can only be good news for the Internet TV industry.