You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!

  • The Intel Apple Deal is a Nothingburger!

    Article: ♫ If you are going to San Francisco, be sure to....-nothing-burger.jpgThe latest Intel rumor that the pro Intel media are flogging is that Intel modems will be in some of the new iPhones. The deal is estimated at around $1B. An “estimated” value of a “rumor” deal is quite funny in itself but let’s take a deeper look at what we are gossiping about here.

    Intel got into the 3G/LTE business after acquiring the Infineon wireless division back in 2011. In 2013 Intel also acquired Fujitsu Wireless. “Rumor” has it that the combined cash outlay for these two deals is an “estimated” $2B. I also heard a rumor that Intel bought smaller companies for its wireless effort including $25M for an LTE company in Dresden which was recently closed down. These acquisitions included hundreds of talented engineers (estimate) some of which no longer work at Intel (rumor).

    The Apple deal was leaked by analyst Gus Richard with Northland Capital Markets and repeated by dozens of “media” outlets:

    Intel’s Modem Wins at Apple: Apple has been evaluating Intel’s model for a while. We now believe that Intel will capture roughly 50% of Apple’s modem business in the upcoming iPhones due to launch September 9th. Further, assuming a 50% share of modem business in the new iPhones we estimate that this win could represent $750M to $1.25B in revenue for Intel in CY16. This is a marque win for Intel and would go a long way to reducing the mobile business losses.

    I’m all for this by the way. Competition is the foundation of the mighty fabless semiconductor ecosystem, so good for Intel, if it is true. I do have some observations worth considering:

    The modem in question is manufactured by TSMC using a 28nm process. Why? Because Infineon and Fujitsu and just about every other wireless chip company uses TSMC at 28nm. In addition to the lengthy design and manufacturing process a modem must go through, further qualifications by the regional carriers are required.

    I’m sure Intel had planned on making a 14nm version of the modem using their second generation FinFET process but I question those plans given that in the not too distant future these modems will be integrated into the SoC. By not too distant future I mean today in some cases and next year in most of the others. I know for a fact that Apple has assembled a talented modem team here in Silicon Valley. Qualcomm and Mediatek already have SoCs with an integrated modem for class 7 LTE (leading edge modems are now class 10).

    Intel is a long time TSMC customer by the way but I highly doubt they get Apple or Qualcomm sized wafer discounts. So you have to ask yourself: Self, how much money is Intel really going to make on this deal? $1B of revenue is nice but not if the margins are significantly lower than the Intel corporate norm. And certainly not if you are getting out of the mobile business altogether which I think Intel should, absolutely.

    Interesting to note, TSMC reported a serious revenue spike in July. Month over month revenue increased 35%, year over year revenue increased 25%, Jan-July year over year increased 28%. Any thoughts on where this revenue spike came from? I will share my observations, opinions, and experience in the comments section and please do the same. I’m in Taiwan next week so I should be able to get a good answer in a couple of days. Sound reasonable?