The most exciting thing that Microsoft CEO Steve Ballmer showed at CES was a version of Windows running on the ARM platform. Microsoft and Intel have been betrothed exclusively to each other since the inception of Windows, but the low-power ARM platform is the one favored by most mobile device players, including Apple.
The fact that Windows can run on non-Intel hardware isn't shocking; most of Windows is written in C and should be portable once recompiled for the ARM architecture. The DEC Alpha and Intel Itanium versions of Windows Server showed it is quite possible for the Windows core operating system to run on other CPUs.
However, desktop Windows has been defined not by source code compatibility, but binary compatibility. The entire ecosystem -- applications, utilities, services, third-party development tools, browser add-ons, and drivers -- assumes a single binary CPU architecture: x86. There is 25 years worth of history and code that basically ensures it would be difficult and confusing to everyone if Microsoft created a binary-incompatible version of Windows for the desktop. That can't be what Microsoft has in mind. Instead, I suspect they plan to salvage as much as they can from their flagship desktop empire to use in a mobile platform, but continue to brand it as Windows.
Yet whatever this new ARM Windows borrows from the existing Windows 7, the user interface will be very different. This cannot simply be a gussied-up Windows 7 on a new type of CPU. Instead, this new UI needs to be optimized for mobile hardware; the entire Windows 7 UI (including Explorer and all the Windows accessories) is not a good fit there.
As for what the actual UI looks like, it's more likely to be something simple but configurable like Microsoft's design for Windows Phone 7. And although the details of the new UI will be important to its success, I don't think those details will affect Microsoft's strategy.
Reading between the lines of what Ballmer showed at CES, I can see a possible strategy the company could follow. The four-point plan is one that Ballmer has enthusiastically expressed before: Developers, developers, developers, developers. These four sets of developers hold the key to Microsoft's success.
The first set of developers targeted by ARM Windows is device makers. The CES demo showed ARM Windows using a standard (but recompiled) Windows printer driver, implying that it wouldn't be that hard for printers, scanners, video, network devices, and other hardware to work on the new platform. For hardware makers, the idea that their existing investment in a Windows driver could provide access to the tablet or even phone market is very attractive, at least in theory.