If Panic and Omni find that all they have done is reduce their revenue by 30% then they won't stay long. I'm not at all sure the extra volume will be enough to offset the cost, especially if the store becomes full. It is going to be interesting.
Hosting costs aren't going to go away, you will still need a web site, you might be able to save a little download bandwidth but that hardly breaks the bank nowadays.
I don't think Apple are being greedy, I'm sure that is a reasonable reflection of their costs, I just wonder how it will add up for developers like Panic or Omnigroup. If their volumes double then it makes sense, if they only increase slightly or not at all then it doesn't.
But that isn't really what you're paying Apple for. You're paying them for the creation of a new market that may (or may not) be better for you. Discovery (formerly marketing) is going to largely be their thing. Customer education will become standardized. Trust - they will teach people that apps are safe and hassle free to install & uninstall. They will teach consumers what an app is.
This will probably amount to far more then an increase/decrease in sales. It will determine which apps succeed and fail. These will be different from the winners and losers in the existing, "natural" market. It will determine aggregate demand for apps. This will also be different. It will affect consumers' expectation for what an app should be. For example, I think that learning curves will need to be reduced. The concept of learning software will be reduced.
A lot of existing software (the big names: MS Office, Adobe CS, etc come to mind immediately) probably won't fit into the app mold comfortably. Developers will have strong incentive to create apps that do. It will be interesting how much software can be "apps."