What will happen if they don't?
You can also turn the question around. What will happen with MS if they don't move to a single platform?
Everybody else has been doing it and some already have achieved this (Java, Linux, chrome/web-OS, LAMP stack, "cloud"). So it's about time MS did too.
But the problem for MS is that they have to do an extremely large technical overhaul, whereas UNIX based kernels and OS'es just scale from small computers to extremely large super clusters. Those designs have the advantage that they can pick a kernel and choose from a number of api's (QT, Java, Python). Microsoft first has to figure out a new kernel design and then try to press all the different legacy Windows api's into a small platform and somehow make it all secure, that's more than just ambitious. And they have to do it all by themselves, while the competition applies open source software development to do the heavy lifting for all the generic stuff end users don't care about.
If MS is smart, they design a new UNIX like kernel, make that open source, port the .NET platform to it and then let developers only use the .NET platform instead of all the legacy api's that still exists in Windows today. That way the applications for the OS will work in the "cloud", on small devices (with a .NET subset) and the OS can be easily ported by anyone to any chip. If they are not smart, well then they have not learned from their mistakes in the past and are in for a huge challenge.
It is clear MS is choosing against general consensus for the alternative path, focussing on devices and single applications instead of RIA/"cloud" (like) applications that already offer the customer experience MS is talking about bringing to customers with their new platform in the future. Next to that, existing "cloud" applications brought via open standards, don't care about what device you are using them on, leaving the customer free to choose whatever device they like (Apple, ChromeOS, webOS, some cheap knockoff with a browser from China).
It will be very interesting to see what philosophy will get the largest adoption in the next 20 years.