When Intel (INTC 0.79%) first mentioned performing semiconductor foundry services, management spoke about it as a "selective" business in which it would take on strategic customers. Furthermore, it was made quite clear under former CEO Paul Otellini's reign that Intel would not take foundry business that would enable a chip competitor. Interestingly, since Otellini (abruptly) retired and former COO, Brian Krzanich, took the helm, those restrictions have been relaxed. At the Nov. 21 investor meeting, Krzanich was quite explicit in that he would be willing to talk to any customer that could make use of Intel's leading-edge process technology (and was willing to pay for it). What is going on here?
This is obviously aimed at Apple
While Krzanich seemed to imply that the company would be willing to discuss business from a direct chip competitor like Qualcomm, it is likely that this "all inclusive" message is aimed at attracting the one foundry customer that would actually matter -- Apple (AAPL +0.86%). Now, one might conceivably view Apple as a chip competitor, but there is a rather subtle nuance here worth exploring.
Since Apple's software ecosystem is all ARM (ARMH +0.00%) based, and since Apple enjoys the flexibility that comes with being able to custom tailor a chip to its own needs (rather than simply requesting certain features from a third-party chip designer), it is unlikely that Apple would ever use an off-the-shelf Intel chip for its iOS devices. Intel, of course, probably wants some way to profit from the immense successes of both the iPad and the iPhone, so if it can't sell the apps processor, why not build Apple's chips? Sure, this business is lower margin than selling an Intel-designed chip, but it would still add meaningfully to the top and bottom lines (and it would starve Samsung/TSMC of this sizable business).
Will Apple bite?
The thing about this kind of deal is that Intel typically wants to be paid a premium for its leading-edge silicon. This means that Intel would probably charge Apple more per wafer than, say, TSMC or Samsung would, and the general purpose foundries are no strangers to eating proverbial dirt for margins. The upside to Apple, though, is that it gets the world's highest-performing, highest-density transistors. Further, since Intel's yields (that is, the number of good chips per silicon wafer) are the best in the business, Apple's costs per die may not be as elevated as the cost per wafer might suggest. All in all, Apple would likely benefit immensely from building chips on Intel's processes.
The real question is: what's stopping Apple? While building chips at Intel seems great on paper, the general purpose foundries have much more experience catering to many different design teams, and have the tool and support infrastructure in place. Intel has spent considerable time building up the infrastructure to enable foundry customers. But, until this arrangement is proven with shipping products from happy foundry customers, Apple probably doesn't want to take that risk at the 14-nanometer generation (although the 10-nanometer generation seems like fair game).
Foolish bottom line
It's unknown whether Intel will get Apple's foundry business in the foreseeable future, but it is clear that Intel will have a substantial manufacturing lead going forward. While using foundries like TSMC and Samsung will allow it to keep pace with other fabless semiconductor players like Qualcomm and NVIDIA in terms of chip performance (which means the iPhone can be very competitive performance-wise with phones based on those companies' silicon), it will have a tough time competing against Intel-based devices on performance alone.
Apple doesn't sell its iPhones based on speeds and feeds, but given the design choices that the company routinely makes in its own chips, it's clear that maximum performance within a given power envelope is the company's top priority. Building at Intel, now that Intel is more open to taking such business, could just give it yet another edge over everybody else.
