Will Apple use an Intel foundry to stamp future Apple A-series chips?

“At microprocessor giant Intel’s developer forum earlier this month, the company disclosed that it will build a future mobile processor for LG Electronics on the former’s upcoming 10-nanometer chip manufacturing technology,” Ashraf Eassa writes for The Motley Fool. “Although the financial impact of this deal to Intel is unlikely to be significant for a number of reasons (more on that in a coming column), this has rekindled the following, seemingly age-old question: ‘Does this mean that Intel will build future A-series processors for Apple?'”

“The A10 chip inside the upcoming iPhone 7 and 7 Plus series of smartphones is expected to be manufactured by Taiwan Semiconductor,” Eassa writes. “The A10 is expected to be manufactured in TSMC’s 10-nanometer manufacturing technology, which should be in mass production a couple of quarters before Intel’s own 10-nanometer technology is in production.”

“Intel’s next real shot at building the processor for the iPhone, then, would be for the A12 chip that will go inside the 2018 iPhone,” Eassa writes. “Intel doesn’t have a proven track record of being able to ramp up tens of millions of a third party’s chip design on a bleeding-edge chip manufacturing technology. TSMC does. Over the long term, Apple may very well choose to go through the work to second-source a design at Intel. However, I wouldn’t expect Apple to risk it with the A12. I could see Apple giving Intel a shot for the A13 for the 2019 iPhone.”

Much more in the full article here.

MacDailyNews Take: Yes, Intel has much to prove first.

15 Comments

    1. It would not, but not because of pulling a trick like that, but because Intel has started to work as a contractor manufacturer only three years ago, and the facilities they offer for that are relatively minor. Apple need more than order of magnitude bigger manufacturing that Intel is able to offer now or in the nearest future. Intel’s latest expansion into ARM core business only as big as making few millions SoCs per year for LG Electronics; other deals/clients are relatively tiny.

  1. Intel is a US company with most of their workforce in the States. If they can do it, Apple should give them the work on the A-series chip. Bringing manufacturing jobs back to the US is only a good thing.

      1. Gotcha, would you prefer that Apple continue buying its A chips from Samsung??? Or sole source from capacity-restrained TSMC?

        Apple needs to diversify its supply, and Intel is as good or better than any other option.

    1. It is indeed a US company, but with major facilities in Israel, Ireland and China, among other places. (Intel does much of the design work for its CPUs and wireless chips in Israel.) but yeah, I’d like to see more of the design and manufacturing in the US.

      IBM also used to do a lot of foundry work here. I’d be happy to see them do such work with Apple, as they’re doing with Enterprise applications.

  2. Why give Intel additional chip business when they have been waiting so long for upgraded chipsets for the Mac market?
    Saying that, by all means, get Intel to develop the capabilities to handle the A-series chips. Apple need alternatives to TMSC if they are going to freeze out Samsung.

  3. The article made one mistake. It would be the A11 that is on TSMCs’ 10nm tech, not this years’ A10.

    But it did get something right. It would be TSMCs’ 7nm tech that would about equal Intel’s 10nm. That’s not entirely correct either, of course, because it’s more complex than that. But both Samsungs 14nm and TSMCs 16nm aren’t really that at all, while Intel’s 14nm is.

    The problem for TSMCs processes, as well as Samsungs’, is that we don’t know when 7nm will be available. While FinFET, an Intel invention, was good for 22nm, it’s required for 14nm. It’s also necessary for 10nm. But, and here’s the problem; It doesn’t work on 7nm. The replacements they’re looking at aren’t working yet either, and as of now, no one knows which one, if any of them, will.

    I suspect that we’ll be stuck on 10nm for longer than most people think. That would be an advantage for Intel for a good three years, until the others catch up, if they do.

    If Apple’s growth gets back on track, that’s over 300 million SoCs a year. It requires an entire chip factory to produce them. If Apple goes to Intel, and I think this modem is a start to that possibility, for several reasons, then it’s very likely that Apple would contribute to the cost of building, or upgrading an intel plant.

    1. While Intel does have enough cash on hand to fund a new wafer plant ($8B~ ish for a new plant) it would be a good use of the Bank of Apple cash resources as well. In exchange for lowering financial risk for Intel, Apple could secure preferential production.

        1. “Need” is such a subjective word. No, Apple does not need Intel. But I don’t think Apple has any intention of going into the wafer business so they need someone to fab their chips. Intel has some unique wafer design and fab technologies that may be useful down the road for increasing performance and reliability.

          1. You mean like the current lackluster chips from Intel, move on Apple. Withhold 20 billion of the buyback money put it to work on something that is important for the long term.

    2. Yes, the A11 is likely to be 10 nm, but no guarantees. It’s absolutely NOT the A10.

      Yes, TSMC’s 16 nm chips have a higher percentage of 16 nm feature elements than Samsung’s 14 nm has of 14 nm feature elements. Intel beats them both. (Just FYI: NO manufacturer has chips that are 100% 14 nm or 16 nm. That’s not how this stuff works.)

      However, there ARE technologies in the labs that work for 7 nm. Test results show it.

      Yes, we’ll be stuck at 10 nm longer than anyone would like. Intel’s new cadence of “shrink, architecture, optimize — shrink, architecture, optimize — etc.” implies a three year cadence. That would imply 2017-2019 for 10 nm then 2020-2022 for 7 nm. I’m expecting that it will be more like 2017-2020 for 10 nm then 2021-2025 for 7 nm with 2026-2031 for 5 nm as the last of the bulk silicon shrinks.

  4. I haven’t thought this through completely, but I do see a potential gain by allowing Intel to produce Apple’s A chips down the road. For phones, it could better integrate the radio chips with the processor. But where it could have a greater impact is in the desktop market. Apple’s A-series chips are quickly moving to “desktop class” and future iterations if this inexpensive chip could be designed expressly for desktops. It seems reasonable that Apple will want to move its computers that way as soon as it is truly feasible from both the hardware and software perspective. That said, Apple is the only major computer maker that has a nearly full fledged operating system running on ARM. The more productivity applications that get written for iOS today, the more likely this shift will take place. I certainly can’t predict when, but the direction seems fairly obvious.

Reader Feedback

This site uses Akismet to reduce spam. Learn how your comment data is processed.