I suppose first I'll talk about the whole Apple vs. Microsoft thing, being an Apple supporter (I try to keep perspective, but it's hard sometimes... ;) ). As to the relative load of work involved due to Apple's control of the hardware, I'm torn. On the one hand, it's mainly going to be an issue of supplying drivers, which as far as I can tell the drivers are written by the 3rd parties. I'll grant that maybe MS might provide significant assistance to the more "important" 3rd parties, and has to deal with making sure all the important drivers work out of the box (after all, who do you think the person on the street would blame if their Windows install went south due to driver issues without ever getting out of setup... some things need to work). Of course, Apple has to deal with some of that too, though probably not to the same extent. But as far as OS design and implementation goes, this seems more like a peripheral issue (sorry) than a core OS development issue, as far as I can tell.
As to the Intel transition, from what I can tell (I'm still PPC bound), it's going well. And it aught to, seeing as Apple did the same thing about ten or so years ago in moving from 68k to PPC. Part of me wishes that instead of "Universal" binaries, Apple were still calling them "Fat" binaries. Probably some PR issues there, but that doesn't stop the nostalgia.
And speaking of the Intel transition brings me to probably my first key point on MS: let's say tomorrow it was announced that the x86 architecture was ceasing development by all parties for some odd reason (not bloody likely, but let's pretend). So, how long and hard of a time do you think MS would have of moving the Windows client codebase over to another architecture, say PPC. Now, I know that there are server versions of Windows which run on Itanium, but I'd assert that it would be a long and hard road. I have a hard time imagining that they'd leave very long without a viable OS (after all, wouldn't want people to by Macs, or to, God forbid, install Linux on their machines, or otherwise let a viable competitor spring up). So I'd guess that they'd try to get their first version for the new platform out ASAP, and I think it would show. I'd guess that it would half work, and be an obvious stopgap. MS would then start talking up their next version of the OS with, among other features (which may or may not show up) would be *actual* compatibility with the new platform.
Thought experiments aside, I think MS's problem is organization. I have no doubt that they have some very smart and capable people working on their OS. They probably work long and hard on it, and on an individual by individual basis do stellar work. But, with a modern commercial OS, that's not enough. You need the people organized as if their in the army preparing to fight a war. And, frankly, having followed the development of Vista from some distance, the impression I got was that this was certainly not the case. Perhaps marketing is running things. Maybe their internal documentation is piss poor (which would help explain why MS is so reluctant to show it to anyone: it's embarassing). Maybe the organization chart of the OS development team, if the lines were pasta colored, would be mistaken for a plate of spaghetti. To be honest, I don't know if these allegations are true or not, but they seem to me close to the mark. The author's statements regarding the state of the source code of Win2k seem to support the idea that perhaps no one is really in charge over there.
That's what I see when I look at Microsoft. And Apple? There's someone definitely in charge over there (he may be a bit crazy, but he seems to be effective), though we'll see how this Leopard delay pans out.