I think they're onto something here.
If there's one thing Microsoft has done time and time again its releasing a product which was "so so" only to improve on it later (IMO usually very well executed improvements, but I will admit to be biased here and there).
Same has applied to Azure IMO. I'm not claiming it didn't have any potential, but initially it sure had a big label stamped on it: "Expensive!". Renting a virtual database with a good storage (2Gb) and fixed bandwith would cost a multiple factor more than renting a /whole/ virtual Linux environment (which, as the reader may know, could easily host /multiple/ virtual databases). Worse: many service providers also support virtual Window servers; Windows Server 2008 - optimized for web usage (so; with IIS, MS SQL, etc)? No problem. More expensive than a Linux server, but still /way/ cheaper than an Azure solution (which only provides the database aspect).
Now, I'm sure there is more to Azure than that. For example, i'm not taking things like redundancy, backups, and all of that into account. But nonetheless it sure did seem that you'd spend a lot of money on something which could be less expensive if only it didn't have the name 'Microsoft' attached to it.
So if MS is really going for the cloud services then I think it makes sense to "start over" this way; drop the name and embed the 'engine' in a broader package. At the very least it is bound to kill any prejudice because most people will most likely eventually approach it as something "new".
And that may very well give them opportunity to approach the consumer market as well.
You got to start somewhere, and I think this could be a step in the right direction.