1. Longhorn is late.
Remember all those articles and opinion pieces online that "Linux's big chance" is now with Longhorn being late and with Longhorn losing features one by one and with many companies claiming that they won't upgrade to Microsoft's new monster? Well, what a better time for Apple to get into the x86 market and steal the thunder from both Longhorn and Linux! If the "time is right" for Linux because of Longhorn's problems, then the time is even more right for Apple! Linux is no real threat on the desktop compared to the Mac OS X experience, while Longhorn hasn't won many hearts either for different reasons we all know.
2. x86-64 is still a virgin market.
There is no OS today for x86-64 that anyone would call "really mature". This is essentially a new platform: a platform on which many existing OSes have to "start all over again." Linux applications still have major problems with x86-64 (weird incompatibilities or even compilation problems), while the newly released WindowsXP-64 has almost no third party drivers for it yet except the very basic stuff. And Microsoft doesn't seem to care to market it either. If Apple were to come over to x86 today, it would be for full 64bit support for their Apple PCs and it would be good enough to compete on fair terms against Linux and Windows. In the 32bit market, Windows XP is simply unbeatable because of its vast hardware compatibility that it enjoys via third parties. Linux tried and it still has a 2-3% of that market. But on x86-64, the market is just different, and Apple has a serious start-off advantage.
Remember, we are not talking about having OSX running on random PCs here, Apple would never get into the "generic PC" market. This market is impossible to support fully; that was one of the reasons BeOS was killed when it moved from PPC to x86 as well. Apple won't make the same mistake. They know better and they have the infrastructure to modify the stock x86 platform to lock-in Mac OS X to their modified PCs only. I am confident that Windows and Linux and FreeBSD would be able to run just fine on these hypothetical Apple PCs, but Mac OS X won't run on your random PC. And it would be better that way (for Apple), as third party hardware manufacturers release hardware faster than anyone could add support for them, even if they had the full hardware specs. And these hardware manufacturers could probably not be persuaded to write brand new drivers for Apple which it will still have a fraction of the market share when on x86, so the logical step for Apple is to "lock" OSX to specific hardware so the user experience remains good.
Regarding software compatibility, I won't be surprised if Apple re-introduces "fat binaries", like NeXTSTEP had. These are binaries that run on both PPC and x86. Of course, lots of third party software will have to be recompiled, but at least it won't be necessary to be re-written or heavily modified as it was in the switch from 68k to PPC in 1993 or from OS9 to OSX in 2001. In other words, the move to x86-64 could be really smooth for users! Emulation does not make much sense as emulating the PPC in "OS-mode" like Apple did with 68k inside PPC is complex and it would be slow [update: this is the only part of this editorial that I didn't predict on target, as there will be emulation *and* fat binary support - Jobs confirmed today the rest of my points in this article]. As for PPC user support, I am sure that Apple would be able to support PPC users for at least 4-5 more years, as they did with 68k support.
For those who claim that Microsoft will never recompile Office for x86-64 for Apple, I can only say that Apple coming to x86 is not bad business for Microsoft initially in terms of "fighting together" Linux. Microsoft has failed to squash the Linux hype but users who go Mac OS X almost never look back. With Apple managing to squash Linux in the x86-64 market, Microsoft will have to fight Apple at a much later future date. And it will be easier for Microsoft to fight an 'enemy' that plays with the same rules as they are rather one that doesn't (open source). My enemy's enemy is my friend, kind of thing... This is a lot like you are getting beaten at both the club and the school, but you give your lunch money to the bullie at the club guy to come and beat the school guy. At the end, you end up with ONE bully instead of two and that's a progress...
With all this in mind, I believe that THIS is the best time for Apple to move to x86. I would argue that the best time was actually last year, but I will give the benefit of the doubt to IBM who seem to have managed to anger Jobs, mostly because the G5 doesn't fit well on laptops rather than because the G5 might not be fast enough. And remember folks, the laptop business, is Apple's business. If IBM can't deliver, it's time to move to someone who can. It makes sense, and the time is right, so why not?