Admittedly I thought that AMD made more sense initially but I think the key might be that Apple isn't really a "computer company" anymore. Laptop and desktop sales aren't nearly as important as iphone, ipod and ipad sales. With that buying AMD may not be as enticing as ARM as there may be more of a future in the small devices. Not sure on the numbers but I would also imagine the margins are much better on the small devices. Small devices are also generally more tightly integrated than laptops/desktops so owning the stack may be advantageous.
With that in mind, why buy ARM? Why not buy something new and really innovate.
We're at the point where there's only one platform that isn't doing multi-architecture development, and even Microsoft has Windows CE / Mobile running on ARM and x86.
With the way things are going, it's conceivable you could see mobiles running MIPS, Power, and half a dozen other technologies. Apple could lead the way if they bought up a couple smaller companies.
I have doubts that Apple is planning to switch to ARM from Intel if for no other reason that ARM is not 64bit and from what I recall it's not in the roadmap. This works fine for small devices but obviously won't work for anything larger. I doubt a PAE-style kernel is a legitimate option.
While I generally enjoy a Cory Doctorow opinion I can't help but to think he is off on this one. The iPad is analogous to a newspaper, both used for consuming content and both gated from seeing how they work on the inside. The Apple IIgs' of today are Arduino's, FPGAs, and the like. Mostly or fully open source and customizable.
