I have read elsewhere that the initial 128 bit support in Win 8 is for a 128 bit file system architecture. It's for hard drives (and removable drives, backup, etc.), and possibly memory addressing. It would not have been an actual kernel / shell ability. (What runs on the processor would not be 128 bit, but only the hard drive format and addressing.)
I do not think we will see 128 bit PC's for a while. 32-bit processing was around for decades, and is still around. Some companies are staying at 32-bit OS'es for now. 64-bit processing is an advantage if and only if the kernel, the operating system, and the application all use 64-bit programming. If you mix, things slow down as addressing is translated. Things would only be worse with 128-bit processing. One of the problems with Apple's G5 processors was that it was touted as the first mainstream 64-bit processor, but everything was run at 32 bits. This made the G5 a glorified (and really freaking HOT) G4 processor. The PowerPC architecture is even worse about 32-bit / 64-bit translation than Intel's (which was outdone by far by AMD's, which replaced the (t)Itanium 64 architecture on the desktop). OSX on a G5 would not be a truly beautiful thing until the kernel, OS, and apps were all 64-bit, so OSX 10.4 and down were 32-bit to maintain compatibility.
I don't think 128-bit architectures would be possible in the consumer or small business market until the field is truly saturated with 64-bit computing.
Mike