My suspicion is that Sharepoint and Infopath will be the next generation of home-brewed company apps. Companies already have them, so they don't need to be buying X, Y or Z to start writing apps to keep their company going.
"Granted selling [products] to other companies seems to be what large organizations do when they die"
I couldn't disagree more. Could you please elaborate?
However, individuals require far more bang for the buck. So the margins tend to be far smaller and you need to sell to large numbers of people constantly to keep up.
At the same time selling to companies tends to focus on boring products so most of your innovative people tend to leave. It's not exactly a binary choice, but companies like AT&T often go from selling to individuals, to making money from companies, to making money from the government. In the end the quality of your competition tends to drop the closer you are to working for the government which is why I think it's a death spiral. Overtime the company becomes unable to deal with change and the next round of innovative companies tend to disrupt them.
PS: Which is not to say old companies like GM can't sell products to customers it's just they stop being able to make much money doing so unless they can leverage a monopoly.
As far as I can tell, all of their server-side software is or will be offered in a Microsoft-hosted-and-managed configuration. And, it won't be trivial to switch back and forth, but it will at least be possible to do so.
Additionally, Microsoft seems to be working with Amazon Web Services since AWS supports Windows Server. If you don't want Microsoft to host your applications then you can do it yourself on AWS or on your own servers.
So, if you are building a web app you can probably ignore Microsoft, but if you are building an OS, or Office product there is still a huge shark swimming in those waters.