The more successful Apple is, the more they need Intel. I think Intel will build the A5 for Apple on 32nm but not 22nm....at least not on 22nm until the core is and Atom Samsung is using a 45nm process. Intel's 32nm process would give a chip half the size, probably much lower and higher performance than the Samsung version. ------
Assuming that Intel can match the foundries on low power process then there is some legs to the idea that Intel becomes the 'fab' for Apple.
However, it would have to be 22nm not 32nm as the A6 could be fabbed at 28nm at Samsung and TSMC.
Finally, until Intel can prove that it can improve Atom Apple would be foolish to commit to it.
If the motivation is to get away from Samsung, why use their 28nm process? From what I hear the TSMC 28nm capacity is tiny and they are having problems with it ala 40nm.
The TriGate transistor is inherently low power which should trump the tricks necessary to get the power down on fine geometry planar transistors.
The Atom will get better and at 22nm TriGate is will outperform an ARM core even if it isn't the most elegant architechture.
There is a reason that Intel is building the world largest semiconductor fab in Oregon and another new one in Arizona. With the 22nm shrink and a flat PC market they certainly don't need this new added capacity. Even Apple won't sop up the huge capacity increase coming at Intel. the shrink and new fabs could support a short term doubling of the top line. I have some guesses on what it could be, but I'm sure that we will all be surprised.
Why would INTC fab AAPL's A6? Isn't the A6 an ARM-based CPU? That would be helping the enemy (ARMH). If I was INTC I would only sell INTC CPU's to AAPL or anyone else for that matter.
That brings up another question I have. It is evident that INTC has the process advantage going forward. So can we expect NOK and even RIMM to jump on the INTC mobile CPU bandwagon shortly? This would give them a potential one-up on AAPL istuff, would it not?