Paul Murphy wrote a recent piece where he argued that Microsoft would create a future open source OS that can be tailored to scale from mobiles / handhelds up to desktop computers as a way to...
...make it easy to publically deal with a BSD legacy, allow them to adopt the Red Hat business model for applications and support, eliminate the need for customer galling anti-piracy programs like WGA, encourage adoption of what would be an appropriate micro-kernel for handheld use, and speed the x86 to PPC transition.
I'm not so sure.
First, I don't agree that Microsoft is on a glide-path to a PPC architecture. The ecosystem around x86 is so large that ditching it would be irresponsibly expensive. Besides, Microsoft is a software company, and the CPU they choose is less important than the software. x86 doesn't get in the way of anything Microsoft wants to do in the desktop / server market, and that's why it's unlikely to be replaced anytime soon.
Second, WGA is galling to geeks, but less so to typical users who press the "Activate" button once and never see it again. Ed Bott has managed to find a few false-positives, but most people who buy a new PC won't have such problems. Their machines were designed to ensure activation success.
Third, why would Microsoft WANT the Red Hat business model for applications and support? What has proven more profitable, sale of software as such or sale of services around a free codebase? What has been a bigger spur to innovation, proprietary software with its associated profits or open source software's "voluntary" contributions (consider the percentage of a typical software stack that is proprietary as you answer that question)?
Will Microsoft do more to use open source products? I think they will (or at least should), assuming the license used by most open source projects is stripped of the entanglements expressly designed to prevent use in proprietary situations. Furthermore, I can see more scope for releasing the source code for pieces of Microsoft software (again, under the right terms). Microsoft, however, will always sell some bits as proprietary software, as that is the model that generates revenue from sale of software as such. That's a good business to be in, and a successful one to boot.
As for making Windows work at mobile / handheld scales, yes, more work can be done in that direction. A more modular Windows would be an ideal. Even without that, however, economics of scale are trending towards making Windows cost-effective in smaller devices. Windows can still be relatively beefy, in other words (with all the application development and hardware compatibility benefits that entails) without being too heavyweight for portable devices of the future.
Modularity, however, does not require that everything be open source. Open source is one way of producing software innovation, and Microsoft should do more to leverage it. That does not, however, imply that it is the ONLY way.