Categories

The PC Calamity

As Intel has improved its products, their demand has decreased. Enormous efforts put into improvements are neither valued nor absorbed. The problem is not with the processors themselves but with the systems within which they are built:

Screen Shot 2013-07-18 at 7-18-11.16.38 AM

PC sales fell again last quarter and the contraction is likely to continue. We received affirmation of this as Intel cut sales and earnings forecasts and the crucial capital spending that creates supply in the longer term.[1]

At the same time, computing device sales have soared.

Screen Shot 2013-07-18 at 7-18-5.07.35 PM

Even excluding Android devices which don’t register with Google’s Play Store (and excluding Windows Phone devices), mobile ARM devices are selling at 2.6 times the rate of Intel-powered devices. Put another way, since the birth of Android nearly as many iOS and Android devices have been sold as PCs.

In terms of install base, a computing category that did not exist six years ago has come to overtake one that has been around for 38 years.

The calamity for Intel has been that they have had no part to play in the new category. Perhaps that is because they had every part to play in the old category.

Notes:

  1. Intel said it was cutting 2013 capital spending to $11 billion. The cut follows a reduction from $13 billion to $12 billion in April. Apple’s budgeted capital spending for fiscal 2013 (ending September) was set at $10 billion.
  • Guest

    Were there really >750M iOS & Android devices sold last quarter? ~2.5B devices in the last year?

    That seems … a bit high.

    • Mercutio

      Yes, the bottom left chart appears to have its y-axis incorrectly labeled. Even if you remove the (thousands) from the title, it still doesn’t match the increments in the cumulative chart.

      • thousands

        It does appear to match without the “(thousands)”. Where are you seeing a clear discrepancy?

      • thousands

        Oh, you mean if you remove it from one, and not the other, I guess. I’m actually not sure what the original comment was talking about. The left chart says that ~200m iOS & Android devices were sold last quarter. Where did the 750m and 2.5b come from?

        Perhaps the post was updated.

      • units

        Agree, the left chart shows ~200M 1Q13, not 750M.

        The increments on the right chart don’t seem to be per quarter, however the two charts roughly match up and the “thousands” unit is correct for both charts i think.

    • http://www.asymco.com Horace Dediu

      There was an error in the original version of the post. It’s been corrected. The ratio is 2.6 to 1 mobile:PC.

      • Jessica Darko

        Chart still shows nearly a million “thousands” of android & ios devices. Over 750M android devices alone, which is not realistic as the only place that claims this many devices is google’s indefensible “activations” number, but nothing in the channel or other avenues supports it.

        This is a real scandal in the making.

  • Jessica Darko

    Please let me know where you’re getting the Samsung Galaxy Tablet sales figures in your chart. To my knowledge these figures are not published anywhere. (The “androids” show they are estimates, and I challenge those estimates since they come from PR agencies rather than authoritative sources. IDC, Gartner, etc, are PR agencies pretending to provide objective analysis.)

    One of the ways this time is like the 1990s is that in the 1990s there was constant propaganda about the success of windows– even when windows was a failure– and about the success of the PC– even when Macs were over %25 of the market– from these same PR agencies that people came to believe that the mac was a small share of the market (less than %5) even though it wasn’t true. (They simply didn’t count most mac sales, only mac sales to corporations that went thru the distributers of PCs, while Apple’s sales channel was separate.)

    Let’s not forget history here… even though many of you may be too young to have been alive during it. :-)

    • https://twitter.com/#!/azulum azulum

      Or put another way, when you’re working with garbage numbers, you can’t really do analysis.

      Would any defense of “Android activation numbers” be considered trollish and fanboyish and indicative of partisan bias? I’ll trust Horace’s numbers and introspection of what those numbers mean.

      One of the ways this time is like the 1990s is that in the 1990s there was constant propaganda about the success of windows– even when windows was a failure– and about the success of the PC– even when Macs were over %25 of the market– from these same PR agencies that people came to believe that the mac was a small share of the market (less than %5) even though it wasn’t true. (They simply didn’t count most mac sales, only mac sales to corporations that went thru the distributers of PCs, while Apple’s sales channel was separate.)

      Not to be goading or confrontational, but would you be able to provide data as evidence to back this up? Because if it is true, that’s a big deal, and it undermines all of the research and anecdotes I’d ever heard about Apple in the Nineties with 3% market share and 90 days from bankruptcy in 1997.

      • Ted_T

        Without giving citations (I’m on the road), I can tell you a couple things: at it’s height, Mac market share was indeed ~25%, but this was in the early ’90s, pre-Windows 95 and likely pre-Windows 3.1.

        Secondly, Apple when Jobs took over in 97 and was or wasn’t at bankruptcy’s doorstep, depending on who you believe, had a higher Mac market share (> %5) than it did later in the Jobs era in the early 2000s after the OS X switch-over, when worldwide market share indeed sunk to 3%. It has been climbing steadily since.

      • JohnDoey

        Market share can be deceiving because the market grows, so market share can fall and yet you are selling way more than previously. Also, the Mac is high-end only, and Jobs only made it moreso after his return. So if eMachines ships millions of junk boxes, Apple’s market share can go down but their profitability and health of the company go way up.

      • Kizedek

        Not sure if marketshare was or wasn’t 25% and when that was (sure it was at the beginning). But, to be fair, it doesn’t necessarily have anything to do with being possibly 90 days from bankruptcy.

        Apple had good marketshare in certain markets, notably education and publishing. Apple probably has less marketshare today, and look at its value now. The difference is the focus, the slim product line, the lack of licensees for the OS, the better than ever integration and quality, the ecosystem, the supply chain innovation, the Stores….

        Almost every single one of these items is exactly what pundits were saying Apple should NOT do because it went against the conventional wisdom: wisdom such as there being no margin in hardware, just software. Yet here we are, with Apple taking most profits in every market it enters and MS et al trying to copy Apple like mad to the extent of restructuring its org and producing hardware in an effort to remain relevant.

  • Pingback: Intel's new CEO and Atom chip face uphill climb to mobility | TNT

  • obarthelemy

    I think the post is missing a few items, and thoroughly mixing up Intel and Wintel:

    1- Intel did miss the boat on Mobile, but they’re working on correcting that mistake. Both their “classic” (Core) and their “mobile” (Atom) Chips/SoCs are getting more competitive in the Mobile space.

    2- For the first time ever, somewhat credible tablets sporting Intel CPUs are available (Asus Fonepad, Samsung Galaxy Tab 3 10.1″)

    3- For the first time ever, somewhat credible smartphones sporting Intel CPUs are available (one Motorola unit, a few 2nd/3rd tier phones)

    4- Android and Intel are *not* antinomic; Android and *Wintel* are. A few Android devices, tablets and smartphones, already run with Intel processors, and an open-source port of Android to x86 exists. If/when Intel SoCs get more competitive, Android OEMs can switch to Intel SoCs with 0 end-user impact (the (rare) exceptions being the Android apps that use native code instead of Dalvik, and possible issues with Intel’s GPUs/drivers, though that’s the case with any new GPU/drivers).

    What I find interesting about what’s happening is the Wintel divorce:
    - Intel is no longer 100% reliant on Windows for end-user OSes, they’re not only running Android, but pushing Tizen (their umpteenth such effort though, and none ever panned out, with the exception of Linux in the server space, but that was aaccidental, as far as Intel are concerned).
    - MS is no longer running only on x86 (discounting their previous PowerPC, MIPS and Itanium flops), and on top of that neither Windows RT nor Windows Phone run on x86 at all (that’s a bit of a given in RT’s case, whose defintion is “Windows for ARM”, but still, MS could have launched a cheaper, lighter, Metro-only, non-PCIX-dependent version of Windows for ARM SoCs).

    Another interesting question is what will happen the other way around; will ARM and/or Android or Chrome OS push into Laptop / Desktop / Server territory ?

    • Kizedek

      “What I find interesting about what’s happening is the Wintel divorce…”

      Isn’t that exactly what Horace is alluding to with his conclusion?

      “The calamity for Intel has been that they have had no part to play in the new category. Perhaps that is because they had every part to play in the old category.”

      As for your four points: whatever these “credible” new mobile devices are, I don’t see anyone lined up at Intel to the tune of 65 million chips per quarter (demand for iPads and iPhones). Thus Intel has cut capital spending. When you see such a customer please let us know.

    • handleym

      Your argument only works if you assume that Intel was successful independent of Windows. I don’t think this is true. Intel succeeded, till they were eventually last man standing, not because they had the best architecture but because they had enough money to create better fabs than anyone else, and to throw engineers at the problem of x86 complexity.

      Those days appear to be ending and the mobile landscape looks very different.

      On the one hand, as I have said a dozen times, the problem with x86 as the basis for a CPU is not that it intrinsically requires (much) more area or (much) more power to provide all that backward compatibility, the problem is the complexity of the design. That complexity means that a desktop CPU takes 7 years+ from start to ship, and the experience with Atom shows that it’s not much better taking 5 years from start to finish. (Desktops run seven staggered design teams in parallel, which is why they have a new chip every year. Atom doesn’t, which is why their second microarchitecture has taken 5 years to follow the first.)

      Compare that to Apple, who starting from nowhere could spin a competitive chip in around two years.

      And this complexity burden is not going away. For Apple (and the rest of ARM, but more so Apple) it just gets easier. ARM8 was specifically designed to toss the complexity from the past that had turned out to a bad idea, and while Apple will doubtless ship a transitional chip that handles both ARM7 and ARM8, they will just as doubtless probably drop that after a year or two and go with pure ARM8. (Other vendors may well be less aggressive, meaning they will pay a complexity tax for a few years longer than Apple.)

      Meanwhile Intel will be struggling along trying to design a CPU that runs ALL of 8086, 80286, 80386, virtual 8086, SMM, and x86-64 and all the rest of it — MMX, x87, every random stupid idea that’s been added over thirty+ years. All useless, all irrelevant to the problem at hand. But Intel (for reasons I cannot understand except pure stubbornness, is unwilling to do the obvious thing — ditch ALL OF IT, all the crap except a stripped down cleaned up x86-64 and AVX which is a good compiler target. They probably will one day, just at the point when it becomes too little too late.)

      On the other hand, the Wintel part of the equation no longer matters. Intel got all that money because people wanted to run Windows, which meant an Intel chip. There was no real opening for competitors, whether PPC, Alpha, MIPS or anyone else.

      Today what people want to run is Android (mostly CPU agnostic) or iOS. The best chip will win based on being the best chip, not based on the good luck of happening to be the one CPU that was targeted by some extremely popular software.

      • obarthelemy

        Aren’t most Linux servers running Intel too ? Isn’t that independent of Windows ? One of the key reasons for Intel’s success is backward compatibility (nowadays called “ecosystem”), but I guess you count ecosystems as valid strong points only when they are fruit-flavoured.

        Intel do have a process advantage, it seems to be holding. They also have an engineering advantage of late (the Pentium IV days are not that far away)… I don’t see how those two aren’t valid advantages ?

        As for Apple starting from scratch. Yep, if you discount buying the IP for a turnkey SoC from ARM, then recruiting a turnkey CPU design team, and doing minor tweaks to the purchased design… They did do that in 2 years. It’s a long way from designing a CPU from scratch, designing new features, primitives, extensions and ASM, a GPU… The work Apple do with their ARM license is very interesting, but still way off what Intel do with their arch. Apple do good work, but their scope is not really comparable.

        As for Intel sticking with obsolete features, the main reason is that there is residual use for them, and that the cost for implementing them is very low, both in terms of design (they’ve already been designed), and in terms of silicon (look at a modern CPU’s die: it’s mainly cache and execution units, the decode unit is piffling.

        As for being “the best” being necessary and sufficient to win in the new CPU market, yeah, shuuuure, because that’s so different from what the PC market was at the beginning of x86, and because “ecosystems” are key when discussing Apple vs Android vs WinPhone, but are irrelevant when disccusing ARM vs x86 vs MIPS vs… And finally, because “being the best” is an easily defined, consensual state which is… (I’m curious ?)

      • handleym

        (a) “Aren’t most Linux servers running Intel too ? Isn’t that independent of Windows ?”
        Are you unfamiliar with what the words “last man standing” mean?

        (b) I have given you specific numbers for the cost of the complexity burden that Intel has chosen to bear. You can claim as much as you like that this burden does not exist, but the design schedule for Intel CPUs suggests otherwise.

        No-one is arguing about “the best” here. This is a site about business. Horace has given us the numbers already suggesting that Intel is in trouble. I’m simply fleshing out those numbers by explaining why Intel isn’t going to pivot so as to make comparably large amounts of money in mobile.

        It didn’t have to be this way. Intel COULD have adopted one of two strategies
        - adopt ARM ISA for mobile and rely on its process advantage OR
        - ship Atom as a clean x86-64 only design without past compatibility.
        Both of these strategies would have allowed for more agility in competing in mobile while (just as important) limiting the rate at which mobile CPUs cannibalize the desktop business. The fact that Intel chose neither strategy strikes me as truly massive incompetence in the face of the obvious, worse even than the criticisms I have of the Win8 business strategy.

      • Walt French

        Your points are good but it’s not clear that Intel could have segmented the market so cleanly without taking a huge hit to the bottom line.

        Today, iPads with maybe $20 CPUs compete against Windows devices on which Intel might make $50 in margin. Intel might have to sacrifice 90% of its per-unit profit, and they would not be able to answer the Lenovo, HP and Dells who mostly sell Windows machines, as to why Intel wants to put them out of business.

        If there’s ANY strategy at Intel other than prayer, it’s that they’ll keep on the high-margin business as long as they can, before entirely revamping their capabilities and pricing across the line. Arguably, if your ideas won’t work (as I think), they could be on that best-option path, and sometime Real Soon Now is the time to cave on the low-end chip pricing, to engage in a bloody price war with the ARM producers.

        Yes, if. And arguably. And bloody.

      • def4

        Why not develop a completely new ISA to compete with ARM, totally unrelated to x86?
        That would have preserved Intel as a vertical player and allowed them to fully exploit their chip design and fabrication expertise.

      • John

        “As for Apple starting from scratch. Yep, if you discount buying the IP for a turnkey SoC from ARM”

        What if you consider that Apple was one of the 3 companies that created ARM in the 1990?

      • obarthelemy

        Indeed: please explain how that is relevant to the present discussion ?

      • def4

        Swift was started from scratch.

      • Walt French

        Methinks you overstate Apple’s hardware chops and overlook their huge software advantage in future efforts.

        Exactly EVERY SINGLE iOS app MUST be produced by Apple’s development system. That system (XCode) today has a switch that targets either ARM or Intel (actually, iOS or Mac). The open-source core of that tech, LLVM, is used to generate programs for a whole range of OTHER CPUs. Apple could totally create a new CPU variant for a value-price iPhone, requiring programmers to make very little adaptation effort for the new chip. Likewise, they could build a brand new chip for iWatch, and have development tools up and running almost instantly.

        As you say, Android has a lot of flexibility but achieves it through a “VM” approach that requires a significant overhead of both CPU speed and memory. If the next wave of devices are MORE compact and battery-constrained, Android will not be competitive there.

        I think we’re past the point of “best” CPU mattering. It’ll all be about how well a given chip can adapt to an ecosystem, whereas before it was always CPU designers who led the software guys.

      • JohnDoey

        The iOS Simulator runs in Intel and apps that are under development run in it, on Intel. So iOS and its apps already run on Intel. They run first on Intel.

      • def4

        The point is that they can very easily run on anything LLVM targets. Intel is one of many equal peers in this picture and must compete on merits with no legacy support advantage.

      • Walt French

        and likewise, if Intel and Apple were to collaborate on a custom design for some new product that had an oddball set of needs, there’d be very few constraints on what the new CPU would have to do. Software is not infinitely flexible, but Apple’s devt system covers all the relevant bases.

      • JohnDoey

        The Intel Mac has about 90% of PC sales at $999 and above. The high-end chips used in Macs have been very profitable for Intel. So they have already succeeded without Windows. For 7–8 years now.

        Also there are a lot of Intel-based servers running Unix and Linux that also are very profitable for Intel.

        Where Intel is struggling is the Windows systems, which have dropped to under $400 ASP and therefore have no money to pay Intel a profit on each sale.

    • Tatil_S

      I would not put too much stock in one Samsung tablet model picking an Intel CPU. 10” is the lowest volume Tab, so it could be a “just in case” project, just like ARM companies courting Win8 RT, despite low sales. Such projects are not always harbinger of more to come. Galaxy Nexus, a flagship smartphone, was designed on a TI platform, yet TI abandoned mobile CPU market less than two years later.

      • obarthelemy

        Indeed, but still, a major Samsung device is not a complete dark horse like Ainol’s MIPS-based tablet last year was.
        Also, the specific reasons why TI exited the mobile ARM space don’t apply for Intel as such.
        It’s actually poetic to see Moore’s law continuated in that form: once computing power has reached “good enough”, CPU prices halve every 18 months

      • ptmmac

        Moore’s Law is dead. Actually it is Denson scaling that is dead, but Intel and Microsoft depended on that scaling to protect their turf. Without increasing processor power at an exponential rate, there is no longer a need to buy a new computer every 3 years. Worse yet Moores law’s claim of dropping costs per transistor is continuing, so good enough computing is becoming more compelling with each new process node.

      • Tatil_S

        Actually, Moore’s law does not say anything regarding CPU power or price or the size of each transistor. It is simply an observation that the number of transistors on a CPU keeps doubling every 18 months. Some of that comes through having smaller transistors and some of it comes from having a larger chip, as the latter requires reducing defect density to keep yields steady. I am not sure if the “law” is dead, yet.

    • JohnDoey

      No, I think it is you who is missing that Horace is talking about what actually happened right now and 3 years past, and you are talking about what may happen in the future if dramatic changes occur and Intel’s dreams come true.

      Intel’s chips are not competitive with ARM in mobile because the mobile software platforms currently require ARM, and mobile makers require SoC customization that ARM licensing provides. Even if Intel reaches performance-per-watt parity (a big if) the mobile industry would have to do a tremendous amount of work to adopt Intel and dramatically change the way they do business.

      Lots of ifs. Nothing that suggests that current trends will change.

      • obarthelemy

        “the mobile software platforms currently require ARM” “to adopt Intel and dramatically change the way they do business”…. What are the Intel-based tablets and phones ? Illusions ?

        Or maybe the current software platforms do not all require ARM, and plopping an Intel SoC in a tablet/phone does not require a complete overhaul of OEMs’ way of doing business ? Why would it need to ?

      • Walt French

        @John, I’m pretty sure Intel doesn’t compete in this space for two reasons: first, the economics are lousy. Their technology requires a LOT of R&D and they can’t amortize it with $2 margins per chip. (TSMC recently said they get $9 for a “high end” CPU, less than half that for a mass-market one.)

        Second, as you say, Intel has designed for total power, not ultra-low power. But check out AnandTech, for example; IIRC, the fact that the CPU gets so much done in an instant or two means that Intel can shut the chip down for longer periods, netting to competitive power used for a given amount of work. It’s just up to the OS writer to handle this; one of Mavericks’s features is that it does. You just have to presume that iOS, ever so much more power-sensitive, has tweaked this as well.

        So Intel is about at the place where they can compete with higher-capabiity ARM designs, such as Tegra, Krait and Snapdragon. The challenge is mostly that they don’t know how to make money while doing it, and yet they can’t afford for competitors to steal away all the industry’s growth.

        And let me be clear: half of mobile is Android that has been compiled to run at least well enough, on Intel. Android apps all run in a VM— a “virtual” CPU that makes them unaware of the actual hardware. The other half run in iOS, which you note is available with a flip of a switch. It’s purely a business decision, profit per unit more than cycles/sec or throughput per watt.

    • markwilcox

      >> Android OEMs can switch to Intel SoCs with 0 end-user impact (the (rare) exceptions being the Android apps that use native code instead of Dalvik

      Check your facts – the rare Android apps that use native code instead of (actually just as well as is enough to cause a problem) would be almost every single game out there. Games being the apps that generate the majority of downloads and income in the Android (and iOS for that matter) ecosystems, that’s quite a big deal with a huge end user impact. Ask Microsoft how much luck they had getting games publishers to develop for Windows Phone 7 (they added native code in WP8). Of course it’s “just” a recompile for x86 plus shipping a fat binary that includes x86 code with ARM code. Games publishers won’t want to do that unless there’s a significant number of devices to target, which is going to be a bit of a chicken and egg problem.

      • Walt French

        Yes, this would be an extra step for those who develop the core engines. (I understand a smaller number of these get used across many games.)

        But when you’re talking several hundred million prospective sales, high-end ones that’d profit from fast CPU and graphics that high-end gamers love, that shouldn’t be too hard. As you say, games are a Big Deal.

      • markwilcox

        There would be extra work for the games engine vendors to optimise for Intel Android builds but Intel could easily pay for that to happen even before it made economic sense otherwise. However, each individual game would also have to be rebuilt and tested for Intel on top of those engines. That’s where the demand chicken and egg problem comes in. Some users of the Unity games engine don’t even bother to target ARM Android devices as well as iOS, even though it’s just a rebuild and test in most cases.

  • Pingback: Intel's new CEO and Atom chip face uphill climb to mobility | TNT

  • Nicholas

    Sadly, I am a poor invest because ARM was a stock recommendation in 1995. Way too early. But, doesn’t that first paragraph read straight out of The Innovator’s Dilemma?

    • Walt French

      You’re describing the Investor’s Innovation Dilemma. Tech business is plagued by high uncertainty, shifting leads and luck driving a company from nowhere to star and back. ‘Twas ever so.

      And therefore, be careful about investing on emotion. High P/Es need to be supported by rapid growth of profits, not just bragging rights. The future is never clear, but it’s ever so much foggier if you don’t do some thoughtful penciling out of future revenues, profits, competition…

  • Walt French

    Ars Technica reviewed some hapless Acer/Win8 system today in a way that really throws the issue into focus.

    Intel has some wonderful technology and has worked enough on power management that it can meet design points for mobile devices. But given the sales data Horace presents above, every Intel-powered mobile device sold would be approximately 1.000 ± 0.001 fewer Intel-powered Windows devices sold. Of course, that means going from a chip that they get $50 or more margin, to one on which they might get $5 of revenue.

    Intel has some very bright people; we know they understood The Innovator’s Dilemma well over a decade ago, and in Apple’s decade-ago approach to them they had an early opportunity to get on the mobile bandwagon. The only explanation that makes any sense is that they were totally blindsided by the explosive growth of mobile, despite having a division in the business and more experience in the development of tech than ANYBODY.

    Obarthelemy notes hopeful initiatives but the story is that there is not yet a viable market for mobile devices that would adequately compensate Intel for its expertise. Most Android apps would run easily on X86, but offer no better an experience than on a $10 ARM chip—and there’s already a dearth of high-function Android apps pushing the envelope for more CPU and even a few million Intel Android tablets wouldn’t change that. Likewise, Microsoft’s Win8 initiative hasn’t produced mobile apps that would leverage the X86. Instead, nVidia, Qualcomm and others are financially encouraged to invest in more ARM horsepower.

    Thus, a broadening umbrella for ARM manufacturers to invest in more powerful chips, with fewer opportunities from any Intel investment in the space. Intel would have to make a major commitment to create CPUs both well-suited to mobile needs AND swallow dramatically smaller margins.

    High-performance machines such as full Windows laptops, desktops and of course servers won’t be at much disadvantage from paying the “Intel tax.” But below $1000, Intel and Windows are deeply uncompetitive a claim and reinforced by the huge SurfaceRT write-down.

    Again, all classic Disruption Theory, maybe exacerbated by the fact that “overserving” the CPU marketplace actually makes for a dramatically worse experience in battery life and size, the two attributes that are most important.

    • Sacto_Joe

      Years later, and the disruption just keeps growing….

      • Walt French

        Yes, but does anybody have any idea why? And when will the Horror end?

      • aftoy

        Bell curve disruption courtesy of Steve Jobs. Wouldn’t
        have taken him long to realize 80% of consumers could do 80% of their computing
        on the device he was prototyping from 2004 till they released the iPhone in
        2007. He definitely was constrained by the ARM CPU’s available circa 2004-07
        which was OK for the rudimentary tasks on the 2007 iPhone. Another 2 years and
        a bigger form factor and better ARM A8 Cortex -voila iPad 2010. No anxiety
        about moving on from Mommy’s apron strings. Microsoft took forever to realize that
        we can get by without a full blown PC.

      • Walt French

        Intel was selling an ARM chip when Apple approached them about what became the iPhone. The chip appears to have had roughly the same power as what Apple ended up with. Intel totally volunteered not to play in the mobile game.

        Intel sez they regret letting bean-counters make the decision. I should say.

      • http://www.asymco.com Horace Dediu

        The reason you play with Apple isn’t because you’ll make money but because you might learn something.

      • Walt French

        So I guess you’re saying Intel looked down their noses at Apple, nothing to learn from those hippie children.

        Yeah, unsurprising. Who needs to be involved in another Newton? But I think somebody must’ve seen it as too much of a long shot. This was before HP sued Intel to force a continuation of the IA64 architecture; how bad would a long shot have been?

        And: aren’t companies in the business of making money? What good does it do them to learn something if they can’t monetize it? It sure seems as if Samsung, to whom Apple turned only after being spurned by Intel, is making a fair amount of money from the purely contractual part of its relationship with Apple, not only the “learned” stuff.

      • http://www.asymco.com Horace Dediu

        My point is that working with Apple should not be decided on ROI but treated as and R&D or even Marketing expense. I’m not sure Samsung really made money from making Apple components. Where they got a return on that investment is in the visibility to Apple’s roadmap and hence their decision to enter smartphones “whole hog”. For Intel to have benefited from the relationship they would have had to have a more flexible business model. The scale of the opportunity might have dawned on them early enough to do something about it.

      • KirkBurgess

        Over the existence of Apple, its possible that its three biggest competitors (Microsoft, Google & Samsung) all got there biggest breaks into new markets by partnering with Apple in some way.

      • aftoy

        Time will tell if Apple can step up after Steve. We could learn from Steve but will we learn anymore from Apple?

      • pk_de_cville

        Steve created an “Apple University” within Apple. He hired wicked smart professors to lead it. The mission he gave it was to teach and preserve Apple’s culture of customer centered design, innovation, and service.

        Soon, we’ll see the first fruit of his endeavor. I’m betting it’s another win for Steve (and the team he assembled – all 70,000!)

      • KirkBurgess

        - iOS 7
        - OS X Mavericks
        - iWork for iCloud
        - iTunes Radio

        These are 4 things apple has released that Steve Jobs very likely had no hand in creating.

        I think Apple is in safe hands.

    • Happy Guy

      Intel didn’t see the perfect storm:
      1. ARM chips were cheap
      2. iOS/Android ecosystem supported ARM chips
      3. Within 3 years there was enough software apps (productivity, games, etc.) built for ARM chips that took 20 years to accumulate for WinTel, between 1980-200, and 99% of the software cost less than $3.99.

      So how does any technology company address a shift in consumer preferences and lifestyle that took less than 3 years to completely change?

      Intel could have copied ARM and license its technology for manufacture by anybody… that’s a huge change in business model however.

      • Jerome

        It’s even worse. Intel DID own and produce ARM chips, they bought StrongARM in 1998. They produced ARM chips (XScale), but saw no value in that business and sold it off in 2006.

      • Happy Guy

        2006, right before an ecosystem was introduced for ARM via iPhone in 2007. Timing timing timing. Hardware is nothing without software ecosystem.

      • Walt French

        Yes, 2006, right AFTER Apple came to Intel looking for a platform. Intel had SOC very similar to what Apple ended up with. If this is about “timing,” it’s that Intel had already decided to unload its tech to another company, and the Apple deal wasn’t plausible enough to justify hanging on to the biz.

        “We ended up not winning it or passing on it, depending on how you want to view it. And the world would have been a lot different if we’d done it,” Otellini told me in a two-hour conversation during his last month at Intel. “The thing you have to remember is that this was before the iPhone was introduced and no one knew what the iPhone would do… At the end of the day, there was a chip that they were interested in that they wanted to pay a certain price for and not a nickel more and that price was below our forecasted cost. I couldn’t see it. It wasn’t one of these things you can make up on volume. And in hindsight, the forecasted cost was wrong and the volume was 100x what anyone thought.”

        (My emphasis.)

        http://www.theatlantic.com/technology/print/2013/05/intel-may-have-lost-the-iphone-battle-but-it-could-still-win-the-mobile-war/275825/

      • pk_de_cville

        4. The impact of multitouch
        5. The impact of Apple’s cartel breaking AT&T exclusive.

    • Mayson

      That ArsTechnica review is really brutal:

      “First 8-inch Windows tablet is a device that shouldn’t exist”
      “Sometimes you can cut costs too far
      “Photographs don’t do justice to how bad this screen is.”

      And those are just the headlines and photo caption…

    • professortom

      Can you expand the hypothesis to cover Macs? Will the Macintosh be Intel’s biggest market in the near future? What are the implications for Intel vis-sis-vis the Mac?

      • Walt French

        In a word, no.

        I don’t see any path to that outcome. Open, albeit skeptically, to a story about the Enterprise throwing away its huge investment in Windows, for systems as yet to be identified.

  • the Ugly Truth

    Incredible. Even more so when you see how the market values MSFT and INTC vis a vis AAPL.

    • jpintobks

      It is remarkable as you said. Markets are upside down. The problem is not only Intel is the whole industry. Google and Microsoft bad numbers yesterday were treated so “mildly” in terms of headlines compared with the harsh and bias way Apple numbers at=re treated. We will see what happen the 23 of July.

      • the Ugly Truth

        yep…both would be trading at half the multiples!

        AAPL, IMHO, will continue to form a base regardless of eps beat on the 23rd. The next tipping point will come in Aug/Sep in time for the next product cycle. Breaking $470 would be great to see. Till then, we are forming base.

  • Conscience_of_a_conservative

    Technology is so unpredictable and wonderful. Seems a decade ago nobody almost nobody saw this coming, yet Wall Street analysts were putting out 20 year growth forecasts for PC makers, Microsoft and Intel. Tech is not Coke and it’s not Nestle.

    • http://www.asymco.com Horace Dediu

      Indeed, and Apple isn’t even tech.

  • Pingback: Imaginando un Microsoft que abandona el mercado de consumo y se centra en la empresa | Descubridores de Tecnologia, Motor y Actualidad

  • Pingback: Imaginando un Microsoft que abandona el mercado de consumo y se centra en la empresa | Civeloo News

  • http://sumocat.blogspot.com Sumocat

    Intel’s predicament mirrors that of Microsoft’s. Both were involved in mobile computing very early (e.g., Intel’s Mobile Internet Device initiative, Microsoft’s Windows Mobile) but were too protective of their PC cash cow to deviate from it. Even today, Intel is still trying to cram x86 into mobiles while Microsoft insists that a unified interface can work across all form factors. The world is changing while they stand still.

  • Pingback: Apple and Android Mobile Devices Devour PC Sales : refertimelylaw.com

  • Pingback: Apple and Android Mobile Devices Devour PC Sales

  • MyCarRunsLinux
    • MyCarRunsLinux

      I’m not as worried about Microsoft as I am about the future of Linux. When PC’s die, Linux as we know it, will bleed uncontrollably. Does Horace ever speculate on the future of Linux?

      • http://www.asymco.com Horace Dediu

        Isn’t Linux at the core of Android?

      • Frode

        “Linux as we know it” with is really called GNU/Windows with two decades of software worth developed for it.

        Linux, as you speak of it is merely a small piece of software written by (essentially) Linus Torvalds that sits between what consumers see and what the hardware sees. It really does not matter what it is, as long as it works.

  • Pingback: The iPad is like the iPod, not the iPhone | stratēchery by Ben Thompson

  • Nex

    It doesn’t matter whether they can maintain the best manufacturing processes, Intel’s business was fundamentally BROKEN once consumers not longer value what they have to offer: x86 and unnecessary performance. Even $50 Pentium laptop/desktop chips are overkill for the general consumer, let alone their $200+ cash cows and the continuous ARM onslaught.