As we all know, Windows has grown bit by bit, byte by byte over the past few decades. As with every system, when it grows so much, we tend to see problems arise in the overall architecture. With Windows, the most notable is probably security; however, if you were on the other end, my guess is that you'd see growing difficulties around integrating new features and expanding upon the current feature-set. Anyone who's maintained a legacy system knows what that's like. It can get pretty bad. Fortunately, things started to change for Windows a few years back. Ever since the invent of .NET, Microsoft has had a vision of modularizing the Windows architecture. Granted, this is simply my opinion based on what I've seen and heard over the years, but now it seems that Gartner is backing me up on it. Nice to see they finally decided to get with the program The only difference is that I've seen a change in direction on Microsoft's part and it seems that Gartner hasn't.
When you go back to the beginnings of Windows Server 2003, you'll see that the plan was to incrementally release portions of Windows one at a time. My guess is that the far-reaching goals of Longhorn caused Microsoft to scale back on their plans. I expect things to change, tho. I've been a big proponent of rewriting Windows along with its many internal components. Honestly, there are probably a lot of Microsoft products that could use a ground-up rewrite. I think we'll see it in time. Plans are already being put together for the next client version of Windows. I suspect we'll see more parts of Windows be built on .NET technologies. Eventually, "native" development will be the wrapper on top of .NET. That might be a hard thing to picture, but it's definitely possible. .NET would have to change and would most definitely need some performance improvements, but that's all possible.