Intel Needs $9bn Mobile Revs To Break-Even

Intel needs mobile chip revenues of $8-10 billion to break-even says Morgan Stanley semiconductor analyst Joseph Moore.Even Qualcomm only has mobile chip sales of $17 billion, so for Intel to get half the revenues Qualcomm gets is a big task.

Especially when Intel does not have, as Qualcomm does, an integrated baseband and apps processor chip.

“$9 billion of revenues simply seems unachievable,” says Moore.

Intel does not see itself integrating baseband and apps processor till the end 2015 or H12016 on 14nm.

And Intel does not see itself making mobile ICs internally until the end of 2015 or H12016 on 14nm.

So, until the end of next year, Intel has to pay TSMC for wafers and then pay customers to design in Intel chips under its ‘contra-revenue’ programme.

While TSMC, GloFo, Samsung all start making mobile chips on 14nm this year.

Moore reckons Intel’s mobile opex at $3.7 billion a year on top of which it has to pay an estimated $50 per chip-set to get customers to use them.

This seems the economics of the mad-house.

Moore reckons Intel will lose a billion a quarter on mobile this year.


Comments

16 comments

  1. Oh yes Intel missed up on a huge opportunity to go outside the company for a CEO when Otellini went. By sticking with an Insider they stuck with a crippling mind-set guaranteeing failure in mobile.

  2. I think we can all agree they didn’t have a clue about mobile. Of course if they’d taken my suggestion of buying a new CEO with his company they’;d be blazing on all cylinders.

    But they didn’t 🙁

  3. You’re right of course, as usual, Mike but in Intel’s case it’s surprising they ever banged the drum about having a process advantage when it came to mobile if 1. Process didn’t deliver any advantage and 2. They weren’t ever going to use the process advantage for mobile.

  4. It relies on how you define “better”/ Certainly almost everyone has now accepted my prediction from 2010 that 20nm, 14nm and 10nm will cost more than each other per transistor, with 28nm being the cost-optimized solution for low performance systems.

    But if you need speed, small die size or lowest power then it is likely one of these nodes will be better in lowering the overall system costs..

    My guess is that an ‘el cheapo’ Chinese tablet doesn’t need that though 🙂

  5. That’s very true, AnotherDavid, but 14nm is, presumably, better than 22nm or 32nm and what has intrigued me all along is why Intel bangs on about wanting to be in mobile, keeps quoting its process excellence as a competitive advantage in attacking mobile, but never puts its mobile ICs on its latest process. It doesn’t make sense to me.

  6. Doesn’t this go back to what does 14nm, 28nm etc mean? I remember quite a few posts on here about this in recent years.
    At the end of the day, Apple and Samsung, which is a huge % of the market, do what they want and the rest its down to the cost of the chips relative to what they do, speed, power etc, with the support / designs that the manufacture gives.

  7. Ah, Yes, thst makes sense, Mike

  8. Qualcomm’s sales for non-integrated basebands outweigh those for integrated. They are bound to if you think about it.

    As for Intel, if you are giving away umpteen billion $ of ICs you make them in your cheapest possible process, which is at Kiryat Gat.

  9. Qualcomm has been ahead of the industry for the past decade Mike and I would bet on it having adopted the correct approach in this case. As a general rule there usually has to be a very good reason for not integrating all the functions of an IC-based solution. As smartphone costs drop integration becomes more important. So why hasn’t Intel ever used a leading edge process for a mobile IC? Is it because iIntel can’t make a leading edge process suitable for mobile? Or is it because the fab managers tell the mobile guys to sod off?

  10. The Intel apps processor is on a process better than Apple or Samsing are currently using. Also since neither Apple nor Samsung have integrated their basebands one has to consider if this is currently the best route. One has to wonder who actually uses the Qualcomm fully integrated solution .. and where ?

  11. Yes Keith, the Morgan Stanley reasoning is: “We believe 2013 losses imply about $3.7 bn of operating expense associated with this business. With a long term gross margin target of 30-50%, this implies annual revenue breakeven at levels 15x the level of revenue reported in 1Q, at $8-10 bn.” And Intel’s Q1 mobile revenues totalled $156 million.

  12. David, can you show us the reasoning behind this statement: “Intel needs mobile chip revenues of $8-10 billion to break-even” ? Why specifically mobile, why not from other sources? Just curious.

  13. Yes indeed jonny wishpants, but at ten years (if you count XScale) and counting it’s taking Intel a hell of a long time to get a boot on the ground.

  14. Yes, but not on an Intel in-house leading edge process, Mike, nor integrated with baseband.

  15. You keep getting this mixed up David. Intel make the apps processors internally already, it’s just the basebands that are made at TSMC.

  16. The important thing is for Intel to get boots-on-the-ground – and that they will do.

Leave a Reply

Your email address will not be published. Required fields are marked *

*