It had long been rumored that Apple (AAPL +4.30%) would finally make the switch from Samsung (NASDAQOTH: SSNLF) to Taiwan Semiconductor (TSM +0.61%) for the manufacture of its custom mobile processors. Chipworks recently confirmed that the Apple A8 sample that it analyzed was built on Taiwan Semiconductor's 20-nanometer manufacturing technology.
However, reports from both IHS and ZDNet suggest that Samsung is still building some of Apple's A8 chips. Taking it a step further, ZDNet reports that Samsung more or less confirmed that Apple's next generation processor will be built at Samsung on its 14-nanometer FinFET manufacturing technology.
This actually fits in well with TSMC's own statements that it will have lower 14/16-nanometer market share than Samsung during 2015.
So, it looks as though Apple and Samsung are still "together" despite a fairly well publicized attempt on Apple's part to move away from the South Korean giant for chip supply.
Samsung seems to really want this business
Samsung has been public in that its 14-nanometer processes offer better chip density than the competing 16-nanometer process variants from Taiwan Semiconductor (some estimates of the relative densities of competing 14/16-nanometer processes can be found here).
While an easy explanation might be that Apple simply chose the best technology for the job, what's actually going on here might be much more complex.
First off, Taiwan Semiconductor runs a highly profitable business selling chips to a very broad array of foundry customers. Samsung, on the other hand, services a much smaller clientele (with Apple widely known as Samsung's largest foundry customer). If Taiwan Semiconductor loses Apple, the former's foundry business won't be crippled.
If Samsung is permanently shut out of Apple, then it will have to be more aggressive in trying to court other longtime Taiwan Semiconductor customers. Unsurprisingly, reports also suggest that Qualcomm will be moving to Samsung at the 14-nanometer node (in fact, there are numerous LinkedIn profiles that suggest Qualcomm engineers are designing for both TSMC and Samsung 16/14-nanometer processes).
The FinFET race looks close
According to an article from EETimes, a Samsung foundry representative was quoted as saying that Samsung and TSMC are "locked in a huge race to 16/14nm production."
My guess is that while both Taiwan Semiconductor and Samsung will use phrases such as "qualified" and "in production" to describe their respective 16/14-nanometer processes, this is small-scale production without the yields to be able to economically support a volume ramp.
This, to me, suggests that the decision for which company will actually win most (if not all) of the Apple foundry business at the 14/16-nanometer node has not actually been made yet; it will probably depend on how both chip manufacturers progress their technological development.
Apple probably can't cut Samsung out as a foundry over the long-haul
While it probably makes some sense for Apple to prefer using a non-Samsung foundry for its chips, it's looking as though over the long-term, Taiwan Semiconductor and Samsung will control the leading-edge of the chip manufacturing market. Apple doesn't have the luxury of ignoring Samsung's foundry offerings.
Intel (INTC +4.45%) could eventually emerge as a viable third option (and for Apple, choosing between Intel and TSMC is probably far more palatable than Samsung and TSMC), but I don't get the impression that Intel's foundry business is mature enough yet to support a customer with Apple's demands.
Foolish bottom line
At the end of the day, even though Apple and Samsung are bitter rivals in the smartphone/tablet space, Samsung is one of the best chip manufacturers out there. While Apple could choose to simply ignore Samsung and put all of its eggs in the Taiwan Semiconductor basket, this probably wouldn't be wise from a risk-management perspective given how crucial having good, cost-effective supply of chips is to Apple's highly profitable device business.
