What would Apple do with PA Semi ?

Ever since Apple acquired PA Semiconductor (if you missed the original news, one of the first reports is here),
I've been pondering what Apple can/will do with PA Semi that would be interesting enough for PA Semi people to stay around. PA Semi arguably has one of the best microprocessor team in the world, but such people are not easy to keep - after all, if you're a good microprocessor designer, it isn't very hard to find another project/company to jump to. Companies are always looking for great engineers, and PA Semi had many. For Apple to have any hope of retaining those people in PA Semi, it has to have some interesting project for them (or, alternatively, large enough money can do the same job, but I doubt Apple is willing to spend much more - good processor designers are not exactly cheap already). Also, buying a microprocessor design house isn't a tactical move - it must have some deep strategic advantage to Apple.

So what new project would be interesting to PA semi folks and also benefit Apple strategically ? I think it must be a new processor for iPhone and iPod. Apple uses various chips for iPhone and iPods - at least from PortalPlayer, SigmaTel and Samsung. I forget the exact figure, but I think the current estimate of iPod and iPhone unit sales for 2008 is somewhere around 70-80M (with iPhone taking 7-8M). I don't know exact sales breakdown between different iPods, but let's say Apple sells 20M iPod Touch and iPhones (which presumably share the same processor). If the development of a processor for iPod costs $100M a year, per-unit cost would be $5. At that price, it's not a slam dunk in terms of economics, but it's within economically viable range. Apple might very well design their own SoC for iPhone specifically, and can reap the economic benefit.

However, another important point is the explosion of smart phones. The processor arms race that happened with PC is only going to happen again with smart phones. It's bound to happen - as people do more and more stuff on their cell phones, people will demand better performance which translates into new capabilities. Part of the reason Apple lost the first PC game was the performance. Apple couldn't quite develop their own processors - their volume wasn't high enough (I don't remember exact numbers, but before iMac, their sales only around 1M a year, if I remember correctly). This time around, the difference is that Apple has enough volume. With that, the economics of having a processor team in-house just might be viable, and provide all the strategic benefit that comes with it.