Updated: Per a comment suggestion, Nick Carr discusses an MS application remoting/virtualization tech acquisition.
Microsoft and VMWare are both now offering free virtualization software. Mostly, people look at this as a Corporate Information Technology topic - or maybe for developers/QA to help with configuration regression testing and the like - and that's not untrue.
But its a very narrow view.
I mentioned the relevance of a new emerging notion - virtualization as an application portability construct, in passing, on another topic (scroll down about halfway).
This levels up to another trend - masking complexity through abstraction, and its an interesting one as we look to the future. We've seen this debate before (many times), though perhaps the essence of it is, I think, best captured by the old RISC vs. CISC arguments.
The (grossly oversimplified) principle of RISC was: for computers, simple equals better. In theory, a smaller instruction set (reduced) means more opportunity for optimization (parallelization, etc), whereas complex instructions make it harder to scale cheaply at faster and faster clock speeds.
If we look at that debate writ slightly larger, we now have technologies like out-of-order instruction scheduling and "application prefetching" built into Vista (to help insta-load increasingly larger applications). Essentially, masking and optimizing complexity through profile-driven optimization - itself another layer of complexity.
I find these OS virtualization technologies to be(philosophically) diametrically opposed to the whole RIA/Applications-as-content model.
Using a Mac or Linux system (or cel phone for that matter), as a desktop replacement, is a lot more practical today than it was even 5 years ago because of increasing application remoting paradigms.
Office and Windows (and their associated complexities) might be the most significant barriers to this, at an Enterprise level. Preserving complexity favors the incumbents - just ask Adobe (after all, if this is true, then why this?)
As we saw with the RISC vs. CISC arguments - usually the incumbent beats the other guy by co-opting enough of the advantages.