back to article Qualcomm, ARM: We thought we had such HOT MODELS...

The history of mobile device processors is that of two companies, Qualcomm and ARM, both with unique licensing models which many envy but few can successfully copy or compete with. As the growth in the wireless world shifts to embedded and larger-screened devices, and the mobile chip balance of power moves towards China and Wi- …

  1. Steve Todd

    I'm not sure that ARM are in trouble here.

    They either licence the IP for their processors, IO interfaces etc, and charge a small fee based on what each individual client is using (certainly too small a fee to be worth most clients developing their own processors) or let the client build their own compatible designs for a lessor fee. They don't force you to use their components (want to use PowerVR rather than Adreo graphics cores, no problems) and the costs are only based on the components used on chip, not the final device.

    Qualcomm are a different matter however. They're going to have to resign themselves to charging per chip, not on the end-user price as the IEE is pushing for.

    1. A Non e-mouse Silver badge

      Re: I'm not sure that ARM are in trouble here.

      Quite. It feels like that author has some grudge against ARM. He says that ARM are currently at the top of their game with mobile phones and then complains that it will take a couple of years for their next strategy to full kick in.

      At the top of your game and planning for when things go bad? I think that's careful, sensible, forward planning.

  2. Charlie Clark Silver badge
    Thumb Down

    Straw man

    The history of mobile device processors is that of two companies, Qualcomm and ARM

    This may be convenient for your argument but is very inaccurate. Lots of companies are involved in the history of mobile processors including, but not limited to, TI, Siemens, Alcatel, ST Thomson, Ericsson, Broadcom. Typical filler from Rethink though Wireless Watch is generally more readable than Faultline.

    A comparison of Qualcomm and ARM based on their history might be interesting: Qualcomm growing out of the US military industrial complex which guarantee both lucrative research projects and, er, help when licensing them. CDMA didn't become the dominant standard in the US and, er, Korea by chance. ARM grew out of the failure of European manufacturer to compete in the high volume PC business.

    In comparison to either Qualcomm or Intel, ARM is a minnow that shouldn't really survive. It's done well because it's viral done right: license ARM for a known quality at an incredibly low price. It's also done well by not diversifying that much. It talks the necessary talk to get journalists to cover it but spends most of its time developing what its vast customer base wants.

    Qualcomm continues to supply the US military, where spending is more than then next 20 countries combined. Like Lockheed Martin, Boeing and the rest, no real danger there as long as this continues.

  3. Alan Denman

    Qualcomms main problem was its patents.

    China really fined them for conflict of interest to prevent that a monopoly that we can all see happening.

    Competition ensures development and that is what helps ARM.

    Just maybe the rest of the world will see that the mobile patents reducing choice down to one in many a country, though that might get corrected to two.

    The second being Intel who have enough patents so as not to be blocked out.

    So 'then there was two' ?

  4. theblackhand

    Disagree with comparision of ARM and Qualcomm

    Qualcomms challenges are arguably of their own making (struggling to deliver a 64-bit SoC that matches the power/performance of previous generation so delivering a reference design instead and an umbrella licencing deal that allowed them to charge a significant mark up on their patent portfolio that has failed a legal challenge). If they can resolve their SoC issues, they will remain a strong player but they will need to get by with less revenue because the Chinese defeat of the umbrella licencing terms will flow through to all of their markets.

    For ARM, I still see a bright future based on the 64-bit reference platform they have delivered - it appears to continue to deliver incremental increases in power and performance and they have room to continue to develop it. With that in mind, they will continue to have a market in the portable/low-end computing space and as they deliver reference designs with more CPU power the market will continue to grow. The growth will come at the expense of the other CPU manufacturers (Intel/AMD) and as long as their licencing terms are considered fair (i.e. allowing mix and match of design components, allowing modifications of designs via a licencing agreement and keeping licence costs reasonable) I can't see them being challenged. If their licencing/pricing structure changes, then there is the potential for MIPS to compete, but the software development inertia would need a fairly substantial change.

  5. Mage

    ARM & Qualcomm

    Not just phones, but routers, washing machines, TVs, tablets with no Mobile, set-boxes, players etc.

    ARM CPUs start at about 50c to compete with 8bit controllers.

    Qualcomm have increasingly concentrated on IP and almost patent trolling. Unlike ARM they sell chips as a method of collecting the royalties.

    Unlike ARM they concentrated for many years on buying up companies and killing the product to strengthen IP on their own chips.

    e,g. Flarion

    Qualcomm and ARM are not at all similar or facing same pressures as Qualcomm is more niche..

  6. All names Taken


    Q: what if smartphones DIE

    A: they will - soon after the killer asteroid hits Earth or after the aliens attack

    Q: When will the smartphone gravy train end?

    A: as soon as something better attracts public attention and people wish to purchase it (like water, food, spacecraft to avoid impaction meteor/asteroid-alien invasion.)

    Alternatively ARM could just design stuff that falls outside of NSA grasping tentacles?

    (that would seem to be a bit of a killer advertising campaign no?)

  7. Anonymous Coward
    Anonymous Coward

    Missing risk for Qualcomm

    Intel. Intel wants to compete in mobile SoCs, and part of that extends to using their purchase of Infineon which competes directly with Qualcomm's cellular chipsets. Intel may have trouble pushing x86 over ARM, but they will be happy to get their foot in the door by replacing Qualcomm chipsets in cell phones that aren't using Qualcomm's ARM SoCs - Apple being the largest such customer.

    There were rumors published last month that Apple would be switching back to having Samsung fab the A9 this fall, because Samsung has / will have their 14nm process ready, once again using Qualcomm's wireless chipset. The timeline further showed Apple's A10 would be fabbed at 10nm in 2016, but this time using an Intel wireless chipset.

    It shows someone else fabbing them, but everyone knows Intel is the only company that will have 10nm ready in 2016, so switching to Intel's wireless chipset would make sense as that might be the 'price' Intel asks to be willing to dirty their fabs by baking a quarter billion ARM SoCs for Apple :)

    1. Charlie Clark Silver badge

      Re: Missing risk for Qualcomm

      but everyone knows Intel is the only company that will have 10nm ready in 2016

      We'll believe that when we see it. Recent geometry shrinks have all been hit by delays.

      Intel only wants in on large margin business. Even the 250 million units you're touting is nothing like enough to cover the costs of a new fab. And switching to making ARM chips would send quite a signal to markets about Intel's faith in its own designs. If Apple were to go with Intel on mobile it would be x86.

      1. Anonymous Coward
        Anonymous Coward

        Re: Missing risk for Qualcomm

        True, but if Intel can't make 10nm work, Samsung and TSMC sure as hell won't.

  8. Christian Berger

    There could be some dangers to ARM

    Just imagine someone brings out a rather powerful, but open platform which can be integrated into a chip. A platform which includes everything you need in a standard way. A platform which can be produced royalty free by everyone.

    Some governments are currently throwing research money at such things. After all you do want to have your own computing systems free from potential backdoors installed by foreign countries.

    The problem with ARM is that it's not a common hardware platform. SoC manufacturers love to customize their chips to provide a vendor lock in. Hardware manufacturers using those SoCs however hate that. It means that going from one SoC to another is near impossible, and you never get true second sources.

    1. Charlie Clark Silver badge

      Re: There could be some dangers to ARM

      The problem with ARM is that it's not a common hardware platform.

      This is pretty much true for the competition apart from perhaps, Intel. But there still isn't a common x86 for mobile platform and Intel is just as interested in lock-in as anyone else. And, until Intel can deliver an SoC of comparable power/performance at a competitive price to ARM, things are unlikely to change. All Intel's recent "wins" have been at a significant expense. Granted, with the money it makes selling x86, Intel can afford to keep subsidising SoC side but is this ever going to make sense in China where they're competing with the likes of Mediatek?

      1. Anonymous Coward
        Anonymous Coward

        Re: There could be some dangers to ARM

        Much to agree with there, however

        " Granted, with the money it makes selling x86, Intel can afford to keep subsidising SoC side "

        isn't that what the Itanium faithful used to say about IA64?

  9. John Savard

    What I Took From the Article

    There's a foundry in Red China capable of making 28 nm chips? Oh noes, we're all DOOMED!

    But I'm surprised that ARM isn't in more trouble already. Never mind Intel using the x86 chip as an Android alternative. Since most Android programs use bytecode rather than native code, other chips more similar to ARM, and thus presumably more suitable to smartphone chips, could be used. Thus, why isn't MIPS entering this market?

    Or, more to the point, some RISC chip vendors are licensing their architectures on generous terms, because they would really like to see them more widely adopted. So why not SPARC smartphones, or PowerPC smartphones?

    1. A Non e-mouse Silver badge

      Re: What I Took From the Article

      ARM was designed from its early days as low power. PowerPC & SPARC were always intended for servers (& desktops) where power is/was less of an issue.

      A more interesting question is: Why didn't MIPS take off? This has been used in low(er) powdered devices.

    2. Charlie Clark Silver badge

      Re: What I Took From the Article

      Since most Android programs use bytecode rather than native code, other chips more similar to ARM

      This is a myth. The (current) move is more and more to native, which is why Intel has been so busy trying to optimise Android for x86 and support cross-compiling.

      SPARC, PowerPC, MIPS, et al. can be easily licensed. Of them only MIPS really has any history in the small device space. IBM, the main force behind PowerPC, has made it clear that it's only interested in the data centre. Oracle, who own SPARC, is interested in selling complete systems. MIPS hasn't been that successful recently because it doesn't offer the right products, which is why Imagination (fuelled by Apple's cash) was able to buy it. It has yet to come up with credible SoC for mobile.

      But ARM's competitive advantage isn't just the CPU, it's the eco-system: switching SoCs while painful is possible within a reasonable period of time ("native" code is essential for drivers) which is what you sometimes see with different revisions of the same device. Switching to a completely different architecture really isn't an option.

    3. John 156

      Re: What I Took From the Article

      "But I'm surprised that ARM isn't in more trouble already. Never mind Intel using the x86 chip as an Android alternative."

      Have a good read of Intel's last results with particular reference to its performance (or lack of it) in mobile before ennunciating an hypothesis like that.

      ARM sold 4.8 billion chips in the mobile space last year; catching up with that from a standing start is a tall order. In fact the only market in which ARM does not have a significant presence is Servers where it would have to compete with your other named venders head on, but currently without the muscle power to make much impact overall.

  10. Anonymous Coward
    Anonymous Coward

    re - "despite a trend for the largest smartphone makers to invest in architectural licences,"

    Err, that so called trend is now down to one here.

    And by the looks of it that Cortex A72 is the Krait of 64 bit, just like the Cortex A12 is the Krait and Swift of 32 bit.

    Observation will tell you that ARM worked alongside Apple and Qualcomm to advance release their brands of A12 and for Apple, the A72 !

    Maybe history will confirm that the A72 came from the co-operation between Apple and ARM.

    If truth be told, the memory penalties of 64 bit will get a mention there too.

  11. Anonymous Coward
    Anonymous Coward

    patents, patents and more patents.

    Looking at the supply line we can see Qualcomm making up to $30 a pop out of the latest LTE chips, even when they do not manufacture the CPU SOC !

    History might tells us that because of this Qualcomm sort of owns ARM.

    Time will tell.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like