I have to say that my limited experience with Salesforce.com is why I think Microsoft isn't going to make it in the hosted CRM field any time soon. First off; they're not using software built for the purpose. They are using a modified version of their regular CRM software. It's a bad idea and almost certainly will add tremendous complications and time to the development and proving processes. Second off; they are late to the party and in order to get anyone as customers who aren't already firmly locked-in to MS products, they are going to have to demonstrate that they are at least as functional and reliable as the established players.
Salesforce.com works, after a fashion. They still have problems related to their basic concepts that are being dealt with, and then there are the problems related to their rapid growth. Microsoft is starting in a hole, so to speak, and that tells me it is going to take them quite a while to get it right enough to stop the cursing on the customer end.
The one place where Microsoft might have a head start is in fail-over, which has caused Salesforce.com no end of complaints. Microsoft is good at redundency. So are others, but MS is definitely in the big leagues here. The biggest complaints Salesforce.com has had have to do with service interruptions due to bungled handovers between machines in trouble and those expected to take up the load at those times. They've made tremendous strides here, even according to their critics and users. If I had to bet, though, I'd put my money on MS having less interruptions of service simply because they do that sort of thing so well.
It will be interesting to watch this unfold. Start an office pool on when MS can have it up and running for all comers. Everyone picks a year and the winner gets the pot. Okay. I'm being cruel. But the fact is that it will take quite a while for them to get up a head of steam and no one should expect different. No matter what Mr. Ballmer says in between bouts of tossing furniture.