A little while back, I wrote a piece looking at the potential options Apple (AAPL +0.30%) has for the manufacture of its next-generation A9 processor. The option that I pegged as least likely was that Apple would build another chip on Taiwan Semiconductor's (TSM +0.57%) 20-nanometer manufacturing technology, particularly as the gains that Apple could see in moving to a 14-nanometer or 16-nanometer technology could be substantial.
However, it's looking less and less likely that either Samsung (NASDAQOTH: SSNLF) or TSMC will be able to deliver in time for a next-generation pair of iPhones.
The TSMC timeline doesn't work
Bernstein Research's Mark Li points out, via Barron's, that since TSMC expects revenue for its 16-nanometer FinFET process to show up in the fourth quarter of 2015, this is far too late for a September iPhone launch. If Apple pushes out the launch of next-generation iPhones, then TSMC could conceivably supply 16-nanometer FinFET chips into those devices, but this seems pretty unlikely, given how fiercely competitive the smartphone market is.
If Apple stays at TSMC for the A9, then it is likely to be a 20-nanometer chip, as has been suggested by both SemiWiki and Digitimes.
Does the Samsung timeline work?
Samsung has repeatedly claimed that it would go into high-volume manufacturing on its 14-nanometer FinFET process by the end of 2014. However, Ultratech (UTEK +0.00%), a chip equipment vendor that claims that its laser spike anneal or LSA tools will be used at "all but one" of the major foundries in the manufacture of 14/16-nanometer chips.
I strongly suspect that the "one" is Taiwan Semiconductor, since if it were either Samsung or GlobalFoundries, then it would need to be both since GlobalFoundries is licensing Samsung's 14-nanometer FinFET process.
Keep in mind that The Wall Street Journal reported that the A8, found inside the iPhone 6 and 6 Plus, entered mass production in the first quarter of 2014 and began shipping to Apple in July. This means that if Samsung is going to produce 14-nanometer chips for Apple, it needs to start producing the chips in the first quarter of 2015.
Can Samsung really get its 14-nanometer yields up from the reported "10%-20%" to something more like 70% over the next five months or so? It seems like a stretch.
Will this be a problem for Apple?
Apple is a company that seems to care very much about the performance and power of its processors. While Apple could improve its architecture and still gain a performance and feature boost with a second-generation 20-nanometer part, the gains would probably be far more dramatic in moving to a 14/16-nanometer FinFET process.
The good news, though, is that the issues with FinFETs don't just hit Apple; they hit the entire fabless semiconductor industry. So it's not as though Qualcomm (QCOM +15.12%) is likely to put out a 14/16-nanometer chip during the time that the iPhone 6s and 6s Plus launch, if indeed Apple keeps with its history or iPhone names.
The one snag, though, is that if these yield issues get resolved by the end of 2015 (and TSMC indicates that it'll start recognizing revenue by the end of 2015), then Android smartphones launching in the early to mid-2016 timeframe could offer meaningfully better performance and power than Apple's 6s and 6s Plus.
The thing to ultimately keep in mind, though, that Apple didn't get to be Apple by making bad technical and business decisions. Whatever solution Apple ultimately opts for will probably be the option that best tries to optimize maximize its sales, margins, and ultimately profits. For investors, that's really all that matters.
