Why the integrated approach to mobile devices is winning
cdixon.org
cdixon.org
First of all it is not at all certain that the integrated approach is winning for smart phones. In fact depending on what definition of winning one uses, it can be easily said that the integrated approach is losing. There are more android devices being sold than iOS devices and the gulf is widening.
Here Dixon uses a bit of sleigh of hand by stating that Android is integrated now because Google bought Motorola. This is not true. Google bought Motorola for the patents, there is no indication that they will change their Android strategy. They still offer the Android software to all other makers at the same terms as they do to Motorola. And furthermore, Motorola is not even the largest Android phone maker. That honor belongs to Samsung and they are very much distinct from Google.
It is also not exactly true that the non-integrated approach always compromises performance. Remember the days of the old pre OSX macs? Those were truly integrated devices with a closed and proprietary architecture from top to bottom. Were they better performing than the PCs? Not even close.
There is an advantage in performance for the non-integrated approach that business people tend to discount. A non-integrated system is much more open and thus it can easily benefit from improvements from unexpected directions. Thus, for PCs we had the improvements like better sound cards, consumer grade video accelerators, and a shitload of improvements of motherboards, memory peripherals, etc. before apple got them. The open PC was ready to take advantage of any brilliant idea anybody in the world ever thought of. For the mac one usually had to get permission from the Apple headquarters first and Apple management were not always perfect or fast.
This is the great advantage of open systems. Their disadvantage is that there are usually problems with interfaces, etc. This was not much of an issue for the relatively large and well powered PC, but may be an issue for cell phones where power, heat dissipation and space are all at a premium.
In general, it seems that the integrated approach wins out in the beginning, when there is no broad marketplace open systems can take advantage of. But after a while, the non-integrated approach wins out. It seems that relatively less-integrated smart phones are slowly but surely beating their more integrated peers. For tablets, however, integration still rules.
If you define winning as profits generated (which is an important definition because sustained profits earn you the right to keep playing, while market share does not), then the integrated approach is definitely winning, as Apple consumes the lion's share of mobile profits (there are several sources but for reference see here: http://www.asymco.com/2012/05/03/the-phone-market-in-2012-a-...).
How things play out in 10 years is anyone's guess, but for now the profits, momentum and sentiment remain firmly on the side of the integrated approach.
Second, when the modular approach was winning the last time around, PC makers weren't the big winners (that was generally a tough business). Instead, Microsoft and Intel were. This time around, that's not Samsung (Galaxy), HTC, LG, Motorola, Sony, etc (profits Asymco looks at) that's Google, Qualcomm, nVidia, TI, Samsung (Exynos), etc (profits Asymco doesn't look at, unsurprisingly, as it isn't obvious how to get them). IIRC, everyone on that second list is generally profitable and very bullish on Android (though none of them breaks out Android specifically, AFAIK). Moreover, Apple's iPhone profits include some amount of return from their SoC and iOS development, so ignoring the second list subtly tilts the scales.
Eg: [Samsung] posted a record net profit for all its divisions of 5.05 trillion won ($A4.43 billion) in the first quarter, thanks largely to strong smartphone sales.[1]
[1] http://www.smh.com.au/digital-life/mobiles/samsung-galaxy-s3...
2) Even if you're comparing the whole company's profits, AAPL still win, they raked in 6 billion USD in profit last quarter.
There is no argument that Apple is making more money. The point is that others are also making excellent mobile profits.
While an integrated approach can squeeze out more performance, does it predict industry structure? Here, it's true that the iPhone's integration greatly aids its performance; and performance is certainly needed in this market.
But (IBM) PC's were never integrated... and performance was certainly in demand in that market (for years and years people would keep buying faster PCs). And, to compare them with Apple's Macs at that time (which were integrated, just like the iPhone), PC's often had better performance per buck than Macs.
Christensen argues that this is because the performance enhancement was more due to particular components (especially the CPU), rather than their integration. Thus, those key components became the focus for improvement. So, the result of Intel improving CPUs (for example) was that the performance of the whole PC increased. This saves the theory... in fact, he turns it around, and points out that Intel CPUs were integrated internally in order to get that performance.
While I do find this very interesting, I get the feeling of post hoc analysis in too many of his examples... and a theory that doesn't predict the future is just not quite compelling for me. :-)
BTW: Christensen says that integrated and component approaches have different strengths, and which one wins depends on what that market needs. If the market is desperate for performance (and will pay a premium for it), then an integrated approach will be more popular and profitable. But if the market doesn't really care about performance, but instead wants something that is customizable, then a component approach will win. Markets change over time. Typically, some feature gets better until it's "good enough" (most people won't pay for more of it). He claims there's also an alternation between the levels of a system: if "integration" wins at one level, then "componentization" wins at the parts that are integrated e.g. PC's were not integrated, but their parts (CPUs, HDD etc) were integrated internally... but the parts used within them switched back to being not integrated.
PCs were in demand for more or less one quality, power (with screen resolution making maybe two). A phone is in demand for a multitude of qualities - size, power, batter life, screen resolution, ergonomics, etc. A PC that was twice the size, twice the power and half the price of its competitor could do well. Phone not so much.
With a multitude of constraints to combine, it's not surprising that software hardware integration happens. Cars are integrated systems more than a hundred years after their invention whereas you can make do buying different components for your house.
Yes, I agree phones are more constrained. But for evaluating the theory, how can you tell how constrained it has to be, to count?
For example, Carmack says that integrated CPUs and GPUs are about twice as fast than as separate modular components (so, for the same power components, an xbox360 can be twice as fast as a PC). And yet, even though graphics power is in demand, the PC market was (and still is) dominated by modular GPUs.
Curiously, Christensen doesn't provide any statistical data across industries (but he did do this for his basic "disruption" idea). I think it's more that, if some quality is in demand, and if that quality can be delivered best by integration, then that might be how it happens. It's a softer theory. I guess the simple answer is that this idea of integrated vs. modular is just one factor of many, so it's not predictive alone.
BTW: surprisingly, cars are not integrated according to Christensen - they are assembled from different components that can be mixed and matched (e.g. the same electrical system can be used in several different models; and different car, same engine; the same chassis can accommodate different versions of each component, because they have bolts in the same places, and use the same interfaces in other respects). This happened because people don't want faster cars (that integrated components are better at delivering); they want safer, more economical cars. But here I have doubts - it seems both those could be better achieved by integration (esp for efficiency).
Yes, I agree phones are more constrained. But for evaluating the theory, how can you tell how constrained it has to be, to count?
That is a good question.
It seems like one simple approach would be to just generalize the big three aspects of phones and cars: Space, Energy and Time. These are kind of the big, orthogonal constraints of our world. You can probably dig deeper and find some others but that's what occurs to me.
People always want cars that are faster, more efficient and fit in a given form-factor. Same with phones. These matter less with desktop machines or say, houses. Laptops are somewhere in-between.
I guess a low-end car might be somewhat less integrated than a higher end car. Etc...
It's really not just about mobile devices, it's about a new platform which is touch and voice based. Its not hard to envision a future where large screens that you can interact with are all over the place, from cars, retail stores, hospitals, replacing poster ads, etc. by touching or speaking with them. These screens will likely be made in different sizes, to accomodate different needs and locations. It's unlikely, in such a scenario, that Apple is going to be manufacturing screens of all sizes to accomodate everyone. It's more likely that a non-integrated approach such as Microsoft's or Google's (or Apple in the unlikely event they change their mind) will deliver on a large scale. The intention of Surface, if you ask me, is Microsoft's way of staying in the game until the above scenario bears out.
Taiwanese manufacturers proved to be willing to become OEM makers for branded products, and a lot of risk has been taken out of the system. Apple couldn't have managed this in the 80s.
The same pattern has been replayed many times over. Even the humble PC was just a toy before Moore's law allowed it to replace minicomputers and eventually take over departmental servers.
How many phone distros are there? How easy are they to install? Why is nobody shocked by these answers?
Turing would be fuckin screaming with stallman; we're regressing and nobody cares.
We all know the old saw that "ideas are a dime a dozen, what matters is execution." What if the strategy (integrated vs. non-integrated) matters a lot less than how well it is executed?
Maybe the reason the "integrated approach" is winning is really that Apple is winning, and the reason Apple is winning is simply that they are executing better than their competitors.
As several others have pointed out, the non-integrated approach kicked Apple's ass in the PC market for years. I think that is simply because Microsoft, Intel, and their hardware partners executed better than Apple. Dell made a better computer than Apple and they did it faster and cheaper too, and marketed it extremely well. Of course they won.
Today with the iPad, it is Apple who is designing a better product, managing the supply chain to make it faster and cheaper, and marketing it extremely well. Their supply chain is so good that [HP could not even get their hands on equivalent parts to build a competitor.[1]
So I think Microsoft can try to go integrated with Surface...it won't matter until they make it better, faster, cheaper than the iPad. Maybe Horace Dediu asked exactly the right question: Who will be Microsoft's Tim Cook?[2]
[1] http://www.theverge.com/2012/6/5/3062611/palm-webos-hp-insid... [2] http://www.asymco.com/2012/06/20/who-will-be-microsofts-tim-...
edit: link formatting