While Intel's (INTC +2.55%) CES keynote focused mostly on wearables and the Internet of Things, CEO Brian Krzanich had many interesting things to say at the JP Morgan Technology Conference. In particular, Krzanich was asked about the motivation behind opening up Intel's semiconductor manufacturing plants. The answer he gave suggests that Apple (AAPL 0.32%), which has a very significant in-house semiconductor team dedicated to building chips for its iPhone/iPad, could be exactly the type of customer Intel wants.
Mea culpa: Apple makes sense
Paul Otellini instilled the notion that Intel should push its own architectures aggressively, so as to eventually drive Apple to simply adopt Intel's chips. However, it has become clear over the last six months that Apple's semiconductor team is exceptionally aggressive and so good at designing chips that Intel will probably never have a shot of getting its own designs in there.
This leaves Intel with two options: Forego Apple's business altogether, letting Samsung or Taiwan Semiconductor have that business, or simply take over the business, as it would likely be accretive to Intel's top and bottom lines. It would take away significant orders from Samsung and TSMC, weakening them, which could eventually hurt their ability to compete as foundries (hurting Intel's more direct chip competitors).
Intel acknowledges that it's better to take the business
When asked about the motivation behind this move, Krzanich had the following to say:
There's going to be a segment of this market that, no matter what, is not going to move to Intel Architecture, it's not going to move to X86. And so as soon as you embrace that, OK, those people aren't going to move, it's OK, I don't have to isolate them. What I can do is embrace that and say, 'how can, for my shareholders, I go make a profit from that space?'
This is interesting because, while the vast majority of the Android space is in fact open to Intel, the only real "closed" ecosystem that would fit this bill would be Apple. While Apple is unlikely to use Intel at 22-nanometer or even 14-nanometer, it seems very possible that Apple would be willing to consider Intel chips at 10-nanometers (where Intel's lead on the rest of the foundries should extend).
Quantifying the impact
Assuming that Intel can get all of Apple's foundry business (approximately 200 million-250 million units per year), and assuming that Intel can get premium pricing for its leading-edge silicon, this could be a deal worth roughly $4 billion-5 billion in sales. Further, Intel expects its foundry efforts to be margin-neutral. The nice thing about this (from a margin perspective) is that Intel's server and PC parts essentially pay for the foundry). This could be worth about $2 billion-$3 billion in additional gross margin dollars.
More importantly, this is business that Intel takes away from Samsung and Taiwan Semiconductor, strengthening its position in the semiconductor industry as a whole, while weakening Intel's major foundry competitors.
Conclusion
Intel wants to own the semiconductor world. If it can get paid design and foundry margin, then it'll take it. If not, then Intel is willing to simply take the foundry margin. Again, the company estimates it will be gross-margin-neutral since Intel will charge a premium for wafers. Customers should be OK with it, as Intel's cost per transistor should be better, according to Krzanich. Foundry could add an entirely new dimension to Intel and help drive the next leg of revenue growth.
