Is Intel's x86 playbook played out?

Is Intel's x86 playbook played out?

Summary: The underlying theory that drives Intel is this: As long as the company keeps making great x86 processors things will be fine. Analysts are beginning to question that logic.

SHARE:
83

The Intel Developer Forum wrapped up this week and the themes sounded familiar: Create better processors and ecosystem, drive devices across all screen sizes and leverage a manufacturing edge.

Intel's taglines---transparent computing for instance---may change, but the underlying theory that drives the chip giant is this: As long as the company keeps making great x86 processors things will be fine.

However, analysts are starting to question Intel's x86 long architecture love affair. These analysts argue that Intel needs a new direction to focus on the post-PC era. New processors won't drive PC or ultrabook sales because people are buying tablets and smartphones, these analysts argue.

The week in IDF: Intel brings voice search to ultrabooks | Intel wants to redefine mobile personal computing via voice, gesture functionality | Haswell: Intel's key for unlocking the post-PC world | Intel SVP: Yes, HTML5 is over-hyped but it will move mobile forward | Intel: We know how to make 10nm chips | Intel ports Android Jelly Bean to x86 phones, but no rollout date set | Intel: Non-volatile memory shift means chips need an overhaul | Inside IDF: Intel shows you the future (photos)

Are the naysayers right? It's unclear, but the argument is worth noting. If the naysayers have Intel pegged, the company is doubling down on reinventing the PC market just as it should be harvesting a mature business and going hog-wild on mobile. Intel's data center strategy can last for a while given that its processors power the cloud---even though ARM-based microservers could be a threat.

Piper Jaffray analyst Auguste Gus Richard outlines the case that Intel is making a wrong turn at a key inflection point. He said:

The Intel Developer Forum was held this week. There was little offered that would change our outlook for the stock. In a post-PC era, we believe CPU performance matters less and the ecosystem and differentiation matter more than ever. It appears, however, the company continues to believe things will be fine if they just build a better CPU. We believe the need for Intel to shift its business model has become obvious. What should Intel do? We would suggest: it improve its system on a chip design methodology, perhaps make an acquisition that allows them to provide OEMs with IP and EDA tools, license ARM and harvest the PC business as it enters a state of decline. At this point, we see Intel as doubling down on improved versions of x86 CPUs, which we continue to believe is the wrong direction.

Richard also called Intel's Haswell chip, which is the successor to Ivy Bridge, a has been. Sure, 10W thermal design points are nice, but the last two Intel processors failed to boost PC demand. Why would Haswell, unveiled at IDF, boost PC demand. Tablets and smartphones made by Apple and Samsung have their own processors. Even Microsoft has an ARM tablet.

intc091412a

In addition, Richard argued that Intel should become a foundry for the industry. After all, Intel's manufacturing prowess can't be replicated easily. Intel could get returns on its manufacturing investment by manufacturing chips for Apple. That talk is heresy at Intel. Meanwhile, licensing ARM is probably heresy too.

Judging from Intel's IDF powwow this week, it's fairly clear that the chip maker isn't exactly consulting Richard on strategy. Intel's David Perlmutter, executive vice president, unveiled Haswell:

I'm extremely proud to show to you and expose to you the next generation of Intel core technology, code name Haswell, which is going to be coming to the world next year. It's based on our 22 nanometer. But the great thing about this one, it was designed with mobility in mind. This is very much to span across the power performance scale anywhere from a sleek tablet to an Ultrabook to eventually a high performing desktop and work station. We have taken the architecture extremely seriously, so we've been able to cut 20x of the item power off the platform level, not just the CPU level, comparing to a Sandy Bridge, which is the second generation core technology.

Will technology buyers care though?

It's possible that tech buyers will care about Haswell---only because the initial Windows 8 systems may be buggy. JMP Securities analyst Alex Gauna noted:

Intel didn't do itself any favors in terms of investor sentiment by trotting out and showcasing Windows 8 systems that clearly were not ready for prime time. Intel is no stranger to demos that go awry, but the main keynote this year struck us as particularly unfortunate and likely to only intensify investor concerns that Intel does not have an answer to the ARM-based computing threat.

Given Intel's sales warning for the third quarter, macroeconomic concerns and ongoing worries about Windows 8 demand, it's not surprising that the company's steady-as-she-goes cadence of new processors is worrying analysts.

Wedbush analyst Betsy Van Hees said in a research note:

While Intel had an impressive lineup of Ultrabooks featured at IDF, we are concerned following industry checks that inventory builds for Ultrabook builds have already taken place with OEMs/ODMs looking for sell through before placing re-orders. Industry checks indicate PC OEMs/ODMs are looking for flat to down 2 to 3% Y/Y growth for PCs suggesting that we could see another quarter of inventory burn in Q4 for MPUs.

That pause in processor sales will hurt Intel in the key holiday shopping season. Given that reality it's no surprise that Intel's naysayer bandwagon is filling up.

Intel CTO Justin Rattner obviously doesn't buy the naysayer argument. His IDF parting shot:

It used to be we'd come out and all we'd talk about was the silicon. As we moved along we talked more and more about platforms and I think today, yesterday and the day before you've come to appreciate the focus we have on creating great experiences across the entire compute continuum. I'm sure we'll enjoy the ride into the future together.

Topics: Processors, Hardware, Intel

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.

Talkback

83 comments
Log in or register to join the discussion
  • Consumers don't care about architecture

    What they do care about is performance and good battery life. So, ARM is more efficient than x86, but x86 is more powerful than ARM. Right now there is a demand for power and efficiency and both companies are moving towards the middle. I wouldn't attempt to predict the future, but I also won't discount x86 architecture.
    General C#
    • Agreed

      And the server world is definitely not post x86 and I don't think it will be for some time. ARM simply doesn't have hardware infrastructure yet to really be a good server chip. A15 quad-core implementations are pretty new. 64bit? It will probably be a while before we see any implementations in quantity that have the interconnect, storage and memory architectures capabilities of Xeon and Opteron. And the cloud is exploding.
      DevGuy_z
    • I think thats very fair.

      I think anyone that starts counting Intel as slowly going down for the count has a big problem for starters.

      There seems to be nothing, not a thing that indicates a small enough, powerful and efficient enough x86 CPU is unsuitable for mobile use. It seems bluntly clear that it should be.

      The whole post PC era is a fictitious phrase used to both explain the downturn in PC sales, which is an erroneous explanation because it should be common knowledge that PC's dont sell so much anymore because PC's dont need to be replaced so often anymore, and the "post PC era" is thrown about as if its proof Microsoft is going to collapse by the Windows haters.

      Its very difficult to fathom why we do not have some more people making the obvious predictions about the future of mobile computing instead of many of the purely ludicrous predictions they do make.

      It seems very obvious that the end goal of mobile computing is to be able to find a way to carry around the power of a desktop in a phone sized piece of hardware. There. I said it. Why don’t more people say it? It’s the truth. And that seems pretty conducive to x86 CPU’s. If they are small enough.

      To be able to do it the way we would like to see it done is at least a ways off yet, but we have taken on bigger challenges than that in the industry.

      You know, when you start asking if Intels x86 platform is played out, your also talking AMD. It’s a pretty massive segment of the IT industry to be making any insinuations at this point that they are idiots for not seeing the writing on the wall when there is actually absolutely nothing to indicate at all that x86 architecture is on its way out.

      But again, we have idiotic predictions being made at all points around ZDNet on a daily basis, largely based on nothing more than the fact the writer either just wants to generate more hits on their story or they are a brand of hater that the prediction would delight them if it was true. Hardly worthwhile reasons for a prediction, but than again it happens as a regular fact around here.
      Cayble
      • Re: There seems to be nothing, not a thing that indicates

        There is one obvious thing: lack of design wins. Practically none of the device makers can be persuaded to use x86 chips.
        ldo17
      • Conducive to x86?

        As long as you don't mind toting around a car battery. Face it even Intel saw the limits of the x86 years ago when it spent a fortune developing Itanium. Big problem, nobody wanted Itanium as there weren't any applications. Had Intel seen the future they would have sunk this Itanium fortune into mobile chips. But when you're a monopoly it's all about preserving what you have...
        GoPower
      • we are all idiotic

        I disagree with you so that makes your views idiotic (to me). You disagree with me so that makes my views idiotic (to you). This type of "circular logic" thinking is why we have a mess in this country now. We all want to label someone that disagrees with us with a disparaging label and that is why we are all idiotic.....except me - only you. So next time you are tempted to assign a derogatory label to someone that disagrees with you, be sure you understand that you are not successfully debating your point. You are simply resorting to name calling and taking cheap shots.
        joneswayne@...
      • The real x86 problem in mobile...

        ...is that mobile device makers are all moving to SOC solutions, for space, weight and power, and also as the 'design space' in which they attempt to differentiate their products.

        If Intel wants into this product space, they not only need a powerful, low power processor, they also need a change of mindset that will allow their customers to incorporate customer proprietary SOC extensions to the Intel offering. And that will represent a huge cultural shift for The sinking 'e'...
        z2217
        • x86/x64 could be shrunk quite a bit

          The problem is that those processors are made for 'higher power' applications right now. Sure, an Atom processor can run Windows 8 (albeit a specially made version) but will it run it well, especially when high intensity or resolution gaming comes into the equation?
          That is a big no at the moment.

          So, no...... x86 will never be big in the smartphones. For tablets, for regular laptops and for desktops? Yes, that chips are perfect for those things as long as low-power consumption x86/x64 chips are used.

          Personally, I don't care if a tablet gets 10 hours of battery life. Most times, if I am near an outlet, I'm going to plug the thing in while I am using it.

          If I am going somewhere that an regular outlet is not an option? I'll bring along a specialized battery charger that fits into a cigarette lighter or car USB port.
          Lerianis10
    • Intel Medfield SoC

      Intel actually already beat ARM in the mobile field technical-wise. The new Intel Atom SoC is able to beat most ARM chips in performance despite only being a single-core. However, it also matches the power consumption of most mid-range ARM chips.

      All Intel really has to do is make a more aggressive push in getting their chips in more mobile devices. Intel Atom is fully capable of defeating ARM. It's a matter of marketing strategy now. Also, not too long ago, Intel showed a demo of an x86 chip running on solar power.
      Generalkidd
  • Intel Low Power CPU

    If Intel have actually cracked this - ARM have zip.
    Master_T[RG]
    • one year

      It will be in latish 2013 that Intel will get there -very low power Quad core SOIC. Intel has an ARM license. We should see TI get bigger on the ARM side.

      What this article doesn't mention is that software is becoming better as well and needing less hardware. Single core Win phones outperform dual core and some quad core phones using other technologies. They went back to a context switching approach.
      mswift@...
      • Re: Single core Win phones outperform dual core ...

        Interesting, then, that Microsoft is completely abandoning the platform that it ran on those single-core Windows phones, and trying to replace it with one requiring two or more cores.
        ldo17
    • Pretty close.

      Intel plans to hit 14nm and then 10nm after that and the power and speed of those monsters will be legendary compared to todays 45nm chips. ARM lacks some core components and instruction set complexity is a silly argument for RISC machines at these complexity scales. It's never been about the instruction set. You save 30% on instruction complexity and lose 30% on more instructions to do what a CISC machine can do. At very small scales it amounts to shifting the processors burden into the memory and that saves power. I've worked at both ends of this spectrum and have seen this first hand. As someone else pointed out, x86 and ARM are converging in the middle and it will become a much more heated battle.
      RyuDarragh
  • The Intel Architecture Has Grown Fat And Complacent

    x86 accrued layer after layer of complexity in an era (the dominance of desktop PCs), and in a market (proprietary Microsoft DOS/Windows software), where backward binary compatibility was of the utmost importance, even more important than power consumption or die size, or even performance. When performance became important (when it was getting thrashed by the RISC chips), it was able to address that by consuming even more power and deploying even more transistors.

    But as a result, it has evolved itself into a corner. Now suddenly there is a new market—mobile devices—growing like Topsy. And as both Intel and Microsoft are discovering to their cost, the ability to run existing Windows software is completely irrelevant to this market. And so the dinosaur is trying to adjust to the fact that the asteroid has hit, while the formerly humble little mammal scurrying around its feet (ARM) is showing signs of inheriting the Earth.
    ldo17
    • Yes

      It's clear to me that Microsoft Windows and buddies (insert Intel and the Dell's of the world) are going the way of those lucrative record labels and recording contracts. Dead dead dead. The world changed, MS isn't cool, their phones can't support any ecosystem, they have no apps and their OS has never been touch orientated but now they are trying it for the first time and hope to swoop up the consumer with high priced products. Interesting sell job. Perhaps Intel will finance some of the marketing for Windows 8 like they are doing for Ultrabooks.
      MisterSavage
      • Actually, Microsoft's phones can support an ecosystem

        And they actually already have one. The problem is that Microsoft was late coming to the 'apps' table and realizing "Oh yeah... people are going to want to put some of these apps on their phones because they make a phone into a portable entertainment powerhouse!"
        Lerianis10
    • Comptition is good

      and should drive better products and innovation, even if it turns the entire market upside down.


      Storied analogies not so much.


      As for Post-PC world: replace every PC in the world with any available mobile device and the world would come to a halt. Mobiles are great PC companions.
      Emacho
      • Competition "is" Good!

        Intel intended to keep their 32-bit and 64-bit product lines separate but all that changed when AMD introduced AMD64 (64-bit extensions to their 32-bit CPUs). This forced Intel to follow suit with their own technology named EM64T. The industry now uses the common label "x86-64". Intel was competing with DEC/Compaq until they acquired Alpha while agreeing to replace Alpha with Itanium. HP had always agreed to replace PA-RISC with Itanium. After introducing technology acquired from DEC-Alpha (most notably the CSI which Intel tweaked then renamed QPI) the resultant hybrid chips were so powerful that some customers began questioning the need for Itanium. It seems that Intel always comes up smelling like a rose.
        neilrieck
        • Thank you for mentioning Alpha

          Alpha was the best innovation (as opposed to evolution) I have ever seen. Intel, like it or lump it, bought it and use the spinoffs of this fantastic chip design. Then Compaq, some two bit IBMPC cheap clone maker, bought out DEC - equivalent to Daewoo buying Rolls-Royce and a lot of knowledge was lost. But that's the way the world works.

          Rather than rail against injustices I choose to hope that the major chip makers continue to do what they do best. ARM should realise that they will never power the web, they should be happy and content in the fact that they power a growing proportion of the devices connecting to the web and they should continue to develop ever faster and more power efficient methods of doing this. Meanwhile, Intel, the only serious player left at the top end (I don't like it, but I'll live with it) should be free to develop the high end server/desktop line and not worry about competition. If Intel try to compete with ARM for the "ho-hum, boring" middle, then I foresee that Moore's Law will finally be broken, and we, as a species trying to evolve, will never get anywhere.
          datax1969
          • AMD can still give Intel a run for it's money

            If it starts getting the prices for it's processors down. Personally, I was looking at a AMD laptop and the only reason I didn't snatch it up was not because of the processor but because after a search? Graphics chip problems with that model.
            Lerianis10