Will It Work?
There's no doubt that porting Windows to another architecture is possible. It's not clear how much backwards compatibility Microsoft intends to provide. When Apple ported OS X to ARM, it didn't provide any backwards compatibility. GUI apps even needed to be rewritten using UIKit instead of AppKit to port them. Apple got away with this by branding the result iOS. Branding something Microsoft Windows 8 will mean that people will expect compatibility with software that ran on Microsoft Windows 7, and explaining to the typical end user about different instruction sets is not likely to make them any happier.
The typical ARM system when Windows 8 is scheduled to shipsometime in 2012will have 2-4 cores and probably 2-4GB of RAM. This is easily enough for an x86 emulator, although it won't perform as well as well as native code. It's also more than enough for .NET code.
At the end of the day, the success or failure of any operating system depends on third parties. MS-DOS would have been a commercial failure without Lotus 1-2-3. When Apple advertises iOS, one of the biggest selling points is the number of third-party applications available. Without applications, Windows on Alpha was less interesting than Windows on an inferior Pentium with lots of apps.
The challenge with Windows 8 on ARM is not technical; it's persuading developers to compile for ARM and test the ARM versions, or to use .NET.