One of them is the abundance of smartphones and dead-cheap computers running any non-Microsoft OS: people will start to see that other manufacturers can make decent OSes. And if you want to make a dead-cheap computer you can't afford to pay Microsoft any money, so you use eg Linux.
The other thing Microsoft must think about is the fastest computers. The kind of computer that appears on top500.org. Linux, Unix and various versions of BSD have a crushingly large market share here, whereas Windows is hardly noticeable. (I can't seem to get any numbers from the site right now, so I am quoting from my memory of the last time I checked).
Since a large portion of supercomputers run something other than Windows, I think things will trickle down from them: you might buy a second-hand supercomputer for your company, and find that it comes with BSD. Or you notice that the computer lab you use to crunch numbers for you will give you a better experience if you use Linux to connect to it. Or some other event makes you realize there are other OSes than Windows. Whatever the reason, some people will sooner or later realize that it is cheaper for them to convert their entire organisation to something other than Windows than to convert their newly-bought servers/supercomputer.
With these two things I think Microsoft will be feeling pressure from the cheapest computers and from the number-crunching monsters. Given enough time they will have to do something to counter these threats, or they will find themselves reduced to one competitor among many. And the transition could potentially be very quick: Altavista disappeared in a matter of years because Google offered something better. Microsoft could end up the same way, though I don't think it could ever be that fast.