Intel's dilemma and the slowly crumbling PC ecosystem

17 thoughts on “Intel's dilemma and the slowly crumbling PC ecosystem”

  1. Between Intel and Microsoft, Intel has the better chance of surviving the transition. It holds a key advantage over its rivals: it owns the process development technologies and manufacturing facilities, whereas its competitors are limited by the ability of companies like TSMC to bring on new process technologies and have to compete with each other (and Apple) for capacity. As Intel gets serious in mobile, I believe you’ll see much faster and lower power processors at better price points than what’s available today. It’s very similar to what happened in the transition from desktop to laptop. Intel was slow to move, but when they did, they eventually dominated the market due to their superior processing technology, processor knowledge and manufacturing prowess.

    1. Greg Borodaty is quite right. The breathtaking Microsoft reversals, surprises and immense new challenges have not occurred overnight, or without management culpability. Those of us who were forced to contend with the legendary Microsoft arrogance in the 1990’s see its current situation with some irony.

  2. As you’re saying, Intel’s biggest obstacles are the company’s DNA and the cost structure. The first one because it means Intel will still focus on doing things the “old way”, and until they adapt, it could be too late.

    And the second, because Intel simply can’t survive in the consumer space (and probably even the server space later, if ARM catches up to them there, too) as long as they don’t adapt their business to be profitable on $20 chips.

    Sure, Atom may be “okay” (it’s not great, certainly not enough to beat the best ARM chips out there), but that Atom business is currently subsidized by the much more profitable Core business. Even if Intel can’t stay competitive with ARM on performance/battery life, it’s going to be very hard for them to stay competitive on price.

    As the ARM chips become faster and faster, Intel will need to either make Atom just as fast or faster (which then presents another problem – what happens to the Core business, if Atom undercuts it?), or make the Core chips as energy efficient as the ARM chips, which even if it were possible, they would still be several years away from achieving it.

    But in both cases, Intel is not going to be profitable. If they continue with Atom, and end up virtually eliminating the Core chips from the consumer market, because they aren’t competitive on price, then Intel *as a company* will have to survive on $20 chips (actually I think Atoms are about twice that right now, so even Atom isn’t competitive on price with ARM chips).

    And if they do get the Core chips to the ARM level in efficiency, that will also mean regressing or stagnating in performance for the “ultra-super-mobile” Core versions, or whatever they will call them. They’ve already showed a proof of concept earlier this year with that dual core 800 Mhz IVB chip (that still had 10-12W TDP I think). And if finally, the ultra-mobile Core chips are just as fast and efficient as ARM chips, they will still be $100+ chips, so again, Intel can’t seriously use that to compete.

    So in the end, what might kill Intel is not the lack of technical expertise in the ultra-efficient chips, but the fact that they will have to lower the prices of their chips so much, that it may not be sustainable for the company – at least without dramatically slashes costs, salaries, and all the “bloat” from the company, so they can be just as “lean” as ARM chip makers.

    1. Lucian,

      Intel actually has an advantage over the other ARM companies in terms of pricing power. The reason is that Intel has its own foundries where as the ARM companies are fabless and have to pay companies like TSMC a lot of money to make their chips. TSMC has margins of 50%, its not cheap.

      Intel not only has their own manufacturing but they are several years ahead in technology. This is a big advantage because it gives Intel a much higher yield on their chips. So, their cost per chip is lower because of the more advanced technology then on top of that Intel gets to keep the foundry margins for themselves instead of paying them to TSMC.

      Also keep in mind that Intel doesn’t just make the chips. They make many different kinds of components. So while the chip itself may only be $20, Intel will also get money from the other components.

      The foundry advantage that Intel has is applicable throughout their product lines. Those factories are not restricted to just one type of chip. They make chips for all kinds of uses. So when you look at market share as a factor driving down costs, you need to look at the entire picture. Its not just smartphones or tablets, it is smartphones + tablets + pcs + servers + embedded devices. That’s a lot of market share for Intel right off the bat even starting at low levels in smartphones and tablets.

  3. Om, There is another trend, computing is going away from computers (things that need booting) towards appliances. People are just sick and tired of being constantly made aware of the short comings of the operating system. They prefer an invisible OS.

  4. Om Malik’s post on Intel is a classic case of “visualizing” a glass half full or half empty. Other competitor bloggers have come to a much different conclusion, as recently as yesterday, based on the identical facts. As one blogger pointed out, Intel and IBM have a prestigious history of reinventing themselves over many decades by plowing money into R&D.. Another problem is seizing on short term Intel results. Peter Drucker is famous for saying, “long-term results cannot be achieved by piling short-term results on short-term results.”

    Longer term, one speaks pessimistically about Intel at their own peril.

  5. I don’t agree that Intel is in any kind of trouble. Intel makes chips and they are way ahead of competitors in their foundry capabilities. They just released a lot of new chips this year that will go into smartphones and tablets. In another year or two they will be able to put out chips using their 14nm technology. So I believe that Intel chips are going to be widely used in the Android market.

    Intel also has a fantastic data center business which is growing very quickly. From what I understand for every six smartphones sold, in aggregate one server is needed to support them.

    Also I would like to mention that even if Intel is never able to get business for Apple’s ios devices they would still have a huge opportunity to provide foundry services for Apple. Intel has the most advanced foundries in the world and Intel could make a lot of money partnering with Apple in this way. The other large foundries like TSMC have margins of 50% and their foundries are not as leading edge as Intel’s.

  6. Intel’s mythical fabrication advantages have largely been irrelevant in the market section that could most benefit from them: Mobile. Node shrinks are supposed to bring advantages in power consumption. At a full node advantage Atom = ARM in performance per watt. That’s kind of bad news since Atom was originally designed to use the spare capacity on older process nodes and maximize the revenue from those investments.

    Also, we’re hitting diminishing returns on node shrinks. We’ve already seen that with the move to 22nm. With both a die shrink and the additon of a Finfet architecture, they barely managed to get half the advantages in speed and wattage that previous shrinks netted them. This may have been due to inefficiencies in the Ivy Bridge chip itself (likely induced by the lack of compitition in the x86 space), or an indication of other issues. From my understanding of the issue ever since 40nm, this has been an issue though with everyone, and can be seen from the modest gains in ARM and GPU architectures as they’ve scaled down, speed and wattage haven’t scaled down in the same ways they used to.

    This means that increasingly the advantage to be had out of die shrinks is a minor performance bump, minor reduction in wattage, and the inevitable price advantages of squeezing more chips onto the same amount of silicon. However, as we move into needing hard radiation to do the lithography at increasingly smaller resolutions, those price advantages are being offset by tremendous R and D costs, as well as ramp up costs.

    We’ll see what Haswell brings. But looking at the market form the outside, it certainly looks like Intel is getting to the point where the competition is able to reach ‘good enough’ faster than Intel can move the CPU performance ceiling. That plus decreased need for more CPU power in the consumer market means that for the first time, Intel is facing a market where it can’t just make a better product that consumers can’t live without. This time they face the very real threat of A) overserving the market, and B) facing disruption from below and on the side. The top of the market may be enough to keep them going, but probably not. Especially if they stumble, or their process advantage fails to net them actual benefits, in which case they could suffer the same fate that Intel inflicted on the Big Iron makers they displaced on their way to market dominance.

  7. Intel is too big. Break it up. This idea will be as welcome as a skunk at a garden party. Break out some of the fabs, like AMD did. Slice the company horizontally, vertically, or both. Some parts won’t survive. This is key. Fear of failure is a great motivator, and Intel can’t see that happening from atop its huge pile of cash.

  8. Intel is in a much better position than Microsoft, they have diversified technologies and are not only dependent on the PC market. Currently the PC/Server market is their cash cow, but they can move into other areas as they have the know-how.

    Microsoft is in a worse position, their mobile efforts is not yielding the results they hoped for, Windows 8 is not the big attraction they bargained on, and Azure not the trumpcard they bargained on.

  9. Intel’s numbers came out yesterday. PCs are only down around 6% while servers are up 8%. The sky is not falling.

    Intel put out 22nm Atom products this year. In another year or two they will have 14nm Atom chips. Intel will be a big player in smartphones and tablets within the next few years.

  10. It seems like we are entering a new era. The convergence of smartphones, tablets and video lliberated from legacy hardware and software have created the neXtWave.
    I go for weeks without my laptop. Wondows did not help their cause with the confused Windows 8 platform. Apple’s laptops seem ridiculously overpriced for what they offer. Google seems to be closer to how our digital lives are unfolding.
    Certsknly there will continue to be a market for laptops. Writing a novel on a smartphone is not practicable. But key market segments are abandoning phone companies, cable companies. The advent of Netflix and other on-demand video sources make older business models seem increasingly irrelevant. Netflix acquired more new clients than HBO. Even Verizon is experimenting with a newer business model.
    The convergence of disparate devices poweted by incompatible operating systems able to combine data from all manner of devices. Cloud storage is eliminating the need to chose one type or format.
    Intel’s market claiim is dwindling. As is Microsoft.
    But as IBM showed jn their latest esrnings reports, good is no longdr good enough.
    Whichever company or companies meet the needs of the convergence evolution will be the future the neXtWave web.

This site uses Akismet to reduce spam. Learn how your comment data is processed.