The processor market is changing too. The old Risc vs Cisc argument hasn't had any real meaning for years, largely thanks to Intel's adoption of the best of both worlds. Intel is committed to moving beyond its chunky NetBurst architecture, the foundation of the hot-running Pentium 4, toward a sleeker, cooler Pentium M-derived future. That will take place late next year, after the company has begun a migration to its 65nm process technology which, it's claimed, may have solved some of the current leakage and power consumption problems that plagued its 90nm process. IBM had similar issues with its own shift to 90nm - one of the reasons why the G5 needs some aggressive cooling systems.
At the same time, AMD has transitioned to its AMD64 architecture, designed from the ground up not only for 64-bit computing, but also multi-core processing. Again, that's an advantage IBM has too, but what IBM lacks are the volumes even AMD is pumping out. With greater volumes come lower prices, and that again makes it harder for Apple to compete.
Apple may go for Intel, it may go for AMD, or use chips from both. Intel is perhaps the most likely, since Apple will undoubtedly be hoping to share in the benefits Dell gets from sticking to a single supplier. If, as is believed, Intel makes it worth Dell's while to stay an Intel-only company, the chip giant is likely to be just as accommodating for Apple. Marketing on the scale Apple works to doesn't come cheap, and any help it can get will be welcomed. Intel gives plenty of help to companies willing to add its logo to their own machines and advertising.
But it may not only be Intel that is driving Apple's CPU shift. At this stage, we don't know what role IBM has played, and the company's own move to become a bespoke chip designer and production foundry may simply no longer favour Apple. Getting a third-party to design your chips for you may be cheaper than doing so yourself, but it's unlikely to be less expensive than buying off-the-shelf parts. Apple may be being forced down the route simply because funding the ongoing evolution of PowerPC is just too expensive to do. While it may have had IBM to share to cost in the past, IBM may no longer be willing to subsidise the development for a chip family that is not only increasingly less central to its own hardware plans but to the future it has in mind for its microelectronics division. IBM even views Cell as a custom chip designed for Sony and Toshiba rather than as a full-scale processor platform it can sell to thousands of hardware developers.
If that's the case, the timing does at least work in Apple's favour. With the shift to the Mac OS X now effectively done, Apple has its most CPU-agnostic OS yet, so it's in the best position it can be to begin transitioning to a different CPU platform. Yes, apps will need recompiling, though Apple may have some clever emulation code up its sleeve to ease the process, as it has done before. That's not to say the shift will be easy for all developers, or even for users, who may find they need to repurchase apps to get the full benefit, but neither is it an impossible move. APIs like Java, Cocoa, Carbon and AppleScript will stay the same, and a lot of code can be left to the compiler to handle. Better compiler technology can bypass many of the limitations of a given architecture.
Then there's the iPod effect. The portable music player has boosted Apple's market recognition to the point where it probably feels it can cope with the loss of the PowerPC zealots and target buyers for whom either the specific brand of CPU doesn't matter, or those who are likely to favour an 'Intel inside' unit. Thanks to Intel's marketing - and pace AMD's superior architecture - there are quite a few folk out there in that position: just look at the chip makers' relative market shares.