Pattern recognition for smartphone investors

I spent the last few days in Hong Kong at a conference discussing the smartphone industry. The participants were mainly investors or investment managers. One of the most frequently asked questions was how to spot investable trends in this notoriously unpredictable sector.

The data I presented did not offer much of an answer. If anything, it showed just how much the industry has changed and how unlikely it is to remain a facsimile of what it is today.

But theory allows us to still make some bold claims. Grounding your investment thesis in pattern recognition rather than extrapolation should be the better strategy. So here are the telltale signs I recommend watching for investable ideas.

  1. In hardware look for what’s not good enough. The smartphone is great. But it’s not good enough for many jobs that we can recognize with little effort. For example, smart devices never seem to have enough battery life. Storage, screen quality and speed are pretty good, but battery life is worse than what voice phones offered years ago. Companies dedicated to improving battery technology or charging technology should be worthy of a bet. Breakthroughs in power saving will be much more valuable than breakthroughs in storage or processing speed or screens.
  2. Pricing needs innovation. Phone pricing is opaque, user unfriendly and often leaves buyers confused and angry. I use the word innovation instead of reduction because it’s not about making prices necessarily lower, but removing complexity and defining value clearly through price. The challenge is not just for devices but for service providers as well. Companies innovating by abstracting pricing complexity will do very well. Interestingly, pricing innovation is not as easily copied as it may seem. It is often coupled with business model innovation and cost restructuring.
  3. Applications currently conform to platform needs but there will come a time when platforms will need to conform to application needs. By applications I mean not just apps but the app/service bundle that will drive consumer purchasing behavior. If we consider social media, entertainment and shopping as applications, it will make sense for those applications to eventually take some control or equity in the platforms. Android is the first sign of this: search and services associated with search and advertising sought control over platforms. Perhaps Facebook, Amazon and Baidu will feel a similar need.
  4. As a new medium, apps will eventually be more valuable than the tools used to render them. The global app brands are probably not yet born, but some day they will be as important as Disney, News Corp and Universal Pictures in their heyday. The App Store is the incubator for these embryonic megabrands.

I realize that these are not specific enough investable ideas but casting your eye on patterns should yield better returns than selecting winners from a set of “current players”.  After all the great returns from smartphones to date have come from bets on entrants (HTC, Apple, RIM) and not from incumbents (Nokia, Microsoft or Motorola).


    • Here's a link to an article by Bill Buxton on the Long Nose Of Innovation. I think it's also a useful model to use in concert with the theory of disruptive innovation.

      A short summary is that the tools needed to innovate are already here and that companies should focus more on refining current technologies than trying to explore new ones. His estimates that it takes 30 years for a completely new technology to go from the lab into the mass market.

      • eyez00

        Excellent article, excellent link, many thanks.

    • I would also add "Product offering rationalization", vendors who manage this properly would gain trust of customers.
      Effectively a corollary of the 2nd point "Pricing needs innovation" but does not often get thought through and hence needs emphasis. When I look at any of the Android vendors there is no clear message around why a specific handset should be chosen. Unending mentions of specifications for many handsets with minor variations only confuses customers like me. Besides the update cycle does not seem disciplined, it is almost as if all non-Apple vendors are churning out every variation possible to see what sticks.

      • davel

        Don't they?

        • Not sure if that was a rhetorical question..

        • davel

          I was referring to "Besides the update cycle does not seem disciplined, it is almost as if all non-Apple vendors are churning out every variation possible to see what sticks. "

          Which I agree with. Yes. It was rhetorical.

    • hakka

      Battery investments are wild – the technology could go anywhere. Lithium titanate? Super capacitors? Or plain old lithium ion with new nanotechnology structures?

      I'd like to think that the industry is bullish, but picking out winners is going to be insanely difficult.

      • HTG

        Perhaps a portfolio approach then? Rather than trying to find the best and 'only' one??

      • Niilolainen

        Interesting viewpoint> diametrically opposed to my own, which is that the battery industry is evolving slowly and incrementally.

        If you believe otherwise I am very genuinely interested to hear why!

        • Iosweeky

          I think you are right on the track record for battery technology being slow in evolution, primarily due to their not being a major market clamoring for a drastically better battery (device makers were happy to build devices around the batteries available).

          However I think that has changed in the last 2 years as the auto industry & consumers finally started to embrace electric vehicles. Present battery tech is unable to supply the auto industry with batteries with enough capacity at a cheap enough price. They NEED a battery revolution, and so the revolution will come.

        • Niilolainen and hakka I think you are both correct.

          When you look at the existing chemistry used in batteries for smartphones and tablets there is Lithium Ion, and Polymer Lithium Ion. Changes for batteries using these chemistries have been incremental certainly for energy density. The risks including safety create cautious approach to changes and a lot of large sample testing is done before adopting changes.

          Here is an excellent resource to learn more about battery technology

          hakka mentioned new battery technologies and different power options. No success making it to production on several that have been trying for a few years now. It always seems some un-foreseen obstacle crops up. So general perception as hakka said is investments are wild. Portfolio investment approach may help but doesn't mean the results could still come up snake eyes as it has so far.

          The batteries in smartphones and tablets account for significant amount of the weight. As well they take up space including separation distance to any radio antennas (required for performance) that limits design flexibility. So there is ample incentive for improvement in energy density required for the power source.

          Other possible improvements are kinetic energy capture including in clothing and solar capture films transparent to the display may help extend battery life or allow for smaller size cells.

          There is certainly a lot of research going on and any significant improvement successfully brought to market will be an extremely rewarding investment.

    • christophe

      given how important the screen is on power consumption, I would not say they are good enough, and would definitely consider screen makers as interesting bet..

      • Niilolainen

        Yes. Definitely. In back/lighting. One of the biggest power consumers in a smartphone. Even worse in a tablet.

        Big benefit of AMOLED is reduced power consumption per sq. mm. See also Quaclomm's investment in Mirasol, other MEMs screen technologies

    • So perhaps not battery companies but instead, renewable energy companies? E.g. Solar cells behind – phone screens etc.

      • Hamranhansenhansen

        I don’t think solar cells belong on the device, but rather on a charger. If you just leave a device in the sun, it will typically overheat and shut down. Also, it could warp components and do permanent damage. And the battery likely really doesn’t like it. So solar cells need to be external. For example, if your backpack had solar cells and a battery and a USB port, you could plug your iPhone into it without putting the device in the sun.

        There is a device that’s already out, the name of which unfortunately escapes me, that you put in your bag or backpack, and as you move around, a piston inside it moves up and down and it generates electricity. It has a USB port on it, where you plug in a devce to charge. Seems like a great solution since it is when you are mobile that you lack power.

        Another idea would be a bike with a USB jack on it that charges your phone as you ride. Or clothing that acted as a solar cell and charged your phone.

        If done right, those are the kinds of things that could essentially make a person into their own charging station.

    • mortjac

      These sentences of yours tells us so clearly about really difficult matters. We should meditate over them, and then act accordingly!
      "…apps will eventually be more valuable than the tools used to render them."
      "The global app brands … will be as important as Disney, News Corp and Universal Pictures in their heyday."
      "The App Store is the incubator for these embryonic megabrands."

      • davel

        Rovio is a private company right?

        • mortjac

          Yes. I also have been wondering if Rovio could be one of these "embryonic megabrands". But they are not anyway near Disney yet, I assume.

        • davel

          Not yet, but they want to make a billion dollar franchise around Angry Birds. Who knows? They might succeed.

      • Hamranhansenhansen

        I think apps are already more valuable.

        iOS is more valuable than an iPhone or iPad or iPod touch hardware, which only exists to frame the system software. And all iOS does is run apps. Period. The T9 buttons of a typical phone are replaced by a 4×5 set of app buttons. You literally just dial up apps. So with no apps, iOS has no value. It’s just a blank screen. And if iOS apps ran on another mobile, it would immediately be very desirable.

        • mortjac

          "And if iOS apps ran on another mobile" is an interesting and new thought. Given that they ran as they do with Apple HW, yes. It would be desirable. Indeed yes.

    • jolin

      Regarding point (2), have you seen giffgaff in the UK ( Simple pricing and bundles of minutes/text/data ('goody bags'). Though it positions itself as a PAYG alternative (and it is), I reckon that with the appropriate goody bag, the service is probably cheaper for most people than a monthly service plan on a contract. And in each month's email statement, they assess your usage and recommend the most cost effective goody bag (or none as in my low-usage case). It will be interesting to see how widely this very simple pricing catches on and whether any competitors emerge.

    • Battery technology is indeed a huge limiting factor these days, but the future is extremely uncertain. There seems to be few incremental improvements to the lithium ion battery these days, suggesting that we've reached some sort of limit with the current technology. I'm sure there will be a breakthrough eventually, but it will be extremely hard to spot. There's also the safety factor. Lithium ion batteries already store an immense amount of energy in a tight space, and a ruptured cell can start a fire. How safe can we get cells with twice the energy density?

      Meanwhile, much of the improvements in battery life is actually coming from greater energy efficiency in processors and smarter software. The battery on the iPad 2, for example, actually holds less charge than the original yet while the device maintains the same battery life. The processor must be more power efficient in order for this equation to balance out.

      In short, I agree that battery life is a huge sticking point in moble, but I think the near-term improvements are going to come from more energy efficient processors, screens, memory, wireless radios, etc.

      • Hamranhansenhansen

        I disagree.

        There was a big breakthrough just recently with bacteria being used to manufacture the battery, and making it 3 times as efficient.

        And only a few years ago, iPad 2 could not be 8 mm thin.

        I think battery progress has just been hidden from us because even as they progress, we keep sticking bigger CPU’s onto them and asking them to attach to even more wireless networks and do even more work per hour. Next to Moore’s Law, battery progress has been slow, but it hasn’t actually been slow.

        Now that we are learning to design around batteries, or at least Apple is, we see “magical” battery life from iPad. Over time, we should actually be able to see the progress of improved raw battery performance in each new iPad.

        • CndnRschr

          Agreed, battery technology has improved significantly – albeit incrementally (a transformative shift is what is holding the electric car back). Apple seems to have recognized this as a strategic advantage to bring in house and the fruits of this are evident from the battery lives of the MacBook Pros, MBA and, to a lesser extent, with iPhone/iPod Touch. The advantage of increased battery is faster adoption of yet to be perfected chips that allow 4G or other functions. Once battery life falls below a day, it is not tolerable by most users. Apple seemed to paint themselves into a corner initially by removing the ability to swap batteries but decision was likely based on the fact that 90+% of people do not carry around spare batteries. They were able to use the integrated battery strategy to increase capacity (no need for room for latches or connectors). There are also significant advances in fuel cells and new metal combinations as well as nanotechnology for increasing storage capacities and decreasing voltage leakage. These are all being driven by a clear need for enhancements in energy storage/recharging* efficiency and cost.

          * With Sanyo Eneloop technology (and others), there is a significant increase in the number of charges that can be made before cell exhaustion as well as charge retention over time. All important aspects of portable reliability.

        • Here's where I'm coming from: The original iPhone, released almost 4 years ago now, had a 1400 mAh battery that looks like this:

          Today, the newest Verizon iPhone 4, has a 1420 mAh battery that looks like this:

          That's a 1.5% increase in capacity after 4 years, and it might all be due to a difference in the size of the battery. Sure, maybe manufacturing processes have improved, and maybe we can increase the number of recharge cycles or reduce how much charge is lost over time. But most people recharge their smartphone every night and get a new one after two years. For that usage pattern, capacity is the most important metric and I don't see the progress on that front.

        • I mentioned the "Long Nose Of Innovation" article because we often hear of technology breakthroughs in the news when in reality the technology is still a long way away from becoming commercialized. I don't know a lot about the technology you are citing and I usually view all news stories with a cynical eye so please forgive my ignorance. But, it reminds me of the 'oil producing bateria' hype that we heard about two years ago.

          If you are an investor, you are looking for technologies that can be commercialized within three years. I just don't see it coming from battery technologies. It's not a funding issue. A lot of funding has been poured into batteries for a long time. The need for better batteries is not a secret.

          For the last 50 years, a lot of research has been focused on the "More." More power, more production, more capacity. But new business models have always been built on the idea of "More with Less." In the last 30 years we went from buying/borrowing movies, to renting movies, to getting movies by mail, to streaming movies. In each of these paradigm shifts, a single VHS/DVD/MP4 could be distributed to many more people than before.

          So if I was an investor looking at technologies that will be big in the mobile computing world, I would narrow my focus to technologies that can provide large performance gains in the next three years. The likely culprits are probably going to come from more efficient chips, screens and software. Incidentally enough, these are the areas where Apple has significantly increased their own investments.

        • Niilolainen

          I think you are wrong. Battery technology is improving slowly relative to the power consumption improvements available from Moore's Law.

          But would be very interested if you could link articles to prove me wrong!

    • A replacement for the Li-Ion battery is definitely in the works in R&D labs and universities all over the world. Someone is going to hit the jackpot eventually, but there's no telling whom that will be. If there's one thing I've learned from reading Horace's posts, it's that disruptions rarely come from incumbents.

      • Niilolainen

        .. has been in the works for many years unfortunately,

        The hydrogen fuel cell has been "5 years away" for at least 15 years. No Moore's Law in chemistry!

    • handleym

      "In hardware look for what’s not good enough. "

      When I look at iPhone4, what I see as the three biggest problems do not match yours. For my purposes, I see the three biggest problems/nice to solve issues as being

      – locked phones in the US. This means that an iPhone is SUBSTANTIALLY less useful than an Android phone for those who travel outside the US a lot.

      – once locking has finally been made obsolete, SIM swapping feels like a totally primitive exercise. Why, when I land in some foreign country, am I buying a physical artifact that I then I have to root around in my phone installing, and which I can easily lose? Why can't I buy some long number that, entered into my phone, validates it with the network? (And, of course, you don;t actually enter the number — you have the number also published as Q-Code or something, so that I can point my phone's camera at it to get the number). Naturally my phone remembers some large collection of these validation codes and automatically switches to them as I travel.

      – optimal GSM performance (voice and data) is great, but all too often we don't see that optimal performance. I've stated the easy solution to this many times — provide antenna diversity in the phone.

      What's sad about this list is that, to truly deal with them well, Apple is constrained by the boundless stupidity of the carriers.
      -I've no idea what the contract terms are that force them to supply ATT with locked phones (or why ATT insists on this stupidity), but, right now, such a contract seems to be in force.
      – Virtual SIMs (and the rest of the infrastructure to make this useful) again requires carriers (now round the world) to show some foresight.
      – While Apple can install two antennas on the GSM iPhone5, and get some improvement on the downlink without involving ATT, they can't take optimal advantage of this technology (and can't improve the uplink) without ATT involvement. Will we finally see progress on this front with ATT's pseudo-4G base stations?

      It's absolutely maddening — and makes one wonder how longer before something serious happens to break this logjam. Nicer batteries et al would be nice, but these three strike me as the single most pressing ways iPhone can be improved.

      • unhinged

        Yes, but would noticing such patterns give an advantage to the investor?

      • Niilolainen

        Your first two points are about pricing (covered in the article).

        Your point about antenna diversity is entirely half baked.

        Old school antenna diversity is pointless unless you can get sufficient de-correlation between received signals, very tricky in a handheld device (and requires parallel receiver chains which cost money and consume power). Antenna diversity is usually a base-site technique. Exception was in PDC where I think it was driven by one DoCoMo dude who also didn't understand it. I don't think any of their suppliers believed in it.

        MIMO on the other hand (which incorporates some diversity type RX techniques) is already incorporated into HSPA+, LTE, 802.11n etc.

        • handleym

          I don't want to argue this, but why are you so sure that MIMO will work where receive diversity will not? They both require de-correlation; they both requireparallel receiver chains. But they solve different problems.MIMO is great- IF the base station supports it AND- if you have a strong enough signal that fades can be compensated by FEC.When this is no longer true, the appropriate thing to do is drop back to a single stream and use receive diversity to try to ensure the error rate is as low as possible.Given that large swathes of the US (and the rest of the world) won't be supporting MIMO for years, and given that better reception is something we all want right away, I don't understand your opposition. Do you deny that VZW phones use two antennas, and that this is for, their non-LTE network, for receive diversity, not MIMO?Diversity strikes me as very much in the same position CDMA was in 20 years ago. The newest students are familiar with and understand it; but we have a huge installed base of engineers who don't really understand it and insist that it can't work, or at least doesn't work well, or at least isn't worth the cost and the hassle.

        • Niilolainen

          Sorry. I take back my half-baked comment.

          I wasn't aware that VZW phones have diversity! It may be that I am in the installed base brigade (been a while since I had an engineering job!).

          I found this references to iPhone 4 CDMA antenna diversity:

          Seems like they use the GPS antenna for CDMA diversity. Better than nothing, but how much?

          How do they get the decorrelation? The DoCoMo phones I am familiar with used field diversity, but with a low-cost coil antenna with -10dB gain relative to the whip and combination only through antenna switching.

          The coil was only used when you actively tried to detune the whip by putting it on a metal surface or something. As implemented then, this arrangement certainly wasn't worth the aggravation, but in the Apple implementation might serve to protect somewhat against "grip of death" type phenomena?

          So on diversity, I am intrigued but still sceptical.

          My understanding of MIMO (as you would probably have guessed by now) is pretty superficial, but my take is that the TX side of things allows you to create decorrelation in time and space via beam forming. ie it is like super-charging your FEC

        • handleym

          I'm afraid I don't have time to go into a seminar on diversity here :-)Let me just catch you up with a few things.- ALL VZW kit now has to support antenna diversity. This is a mandate from ZVW, and I suspect it is one of the reasons that people routinely praise their network as being far more reliable. As far as I can, this is pure receive diversity and VZW is not using Alamouti coding or anything fancier to further improve their network.- Tests have shown that in urban environments, you get pretty much all the benefits of de-correlation at a substantial fraction of a wavelength separation. I forget the exact number — 15% or 25% — but it was remarkably small. I would not have thought this (and I suspect no-one else did) until the tests were carried out. But, honestly, who can say they have an ACCURATE intuitive feel for the scattering environment of the average urban environment.- You can see the geometry of the iPhone4 CDMA antennas by looking at the ifixit teardown. They get some benefit not just from spatial separation but also from being at different orientations in space.- Beam forming is different from (and somewhat orthogonal to) diversity. Beam forming is a way to increase the MEAN power received by a device. Diversity is a way to reduce the VARIATION in the power received by a device.To understand diversity, you first need to understand the nature of the variation in the power of a signal seen by a device. This signal (because of mult-ipath reflections and interference) varies by orders of magnitude over spatial scales of cm, and time scales of ms. The simplest form of receive diversity simply has two antennas with two independent RF chains, and some logic to decide which antenna to use.Assume we have a signal whose MEAN is strong, or at least acceptable, but which, with probability P, is weak because of fading (ie multipath interference). Now, with what are essentially two independent signals, the probability of the signal being too weak is P^2 which is much much less than P.Once this concept is understood, you might ask the question: can we not do a better job of combining these two signals than just choosing the larger one? This requires you to model the state of the “channel” (the signal with all its multipath reflections) to each antenna, as a 1×2 complex matrix. By considering the properties of the matrix, you can indeed do better and combine the two signals to get better results than either on individually.The next step is to consider this matrix idea and ask what happens if you have TWO transmit antennas sending to two receive antennas. There are now multiple things you can do:One is to try for more robust transmission, an extension of the receive diversity idea. The simplest form of this is called Alamouti coding, and more generally it is called space-time coding. This sends linear combinations of the next two (or three, or four) symbols in the stream out different antennas, and repeats this, but with different symbols at different times. Conceptually (this is not quite right) send symbols A+B out antenna 1, !-B out antenna 2, then the next period swap these so antenna 1 gets A-B, and antenna 2 gets A+B. This has even better results (in terms of resistance to fading) than the simple antenna diversity I described above. So essentially this technique and its generalizations are for use in environments where the signal is weak enough that when fading occurs it's a disaster and you lose your information completely.The alternative thing you can do is to send DIFFERENT streams of data (at the full data rate) out the two (or more) antennas available, and have the receiver (again with two or more antennas) separate them. This seems like absolute magic when you first hear about it; but essentially it requires having a reasonably accurate model of the channel (now a 2×2 complex matrix, updated on a ms timescale) which you can use (through essentially a kind of fancy matrix inversion) to multiply against the data streams you receive from your two receive antennas, to get the two streams that were sent.NOTE For this to work, you are no longer relying on redundancy in your two antennas to cope with fading — you now require your two antennas both to provide a clean, error-free signal.But to really understand how this works, you need to be willing to look at both the relevant linear algebra and statistics/probability. If you really do want to learn about it, MIT Open Courseware has a number of good graduate level courses on these issues. But, those courses are every bit what you think they are — they are graduate level courses in EE from MIT — they are not trivial to understand. Alternatively, you can just do a google search on something like “antenna diversity PDF” or “MIMO PDF”, look through what comes up, and pick up as much detail as you wish to know by skimming the results.Thus we have two complementary systems.a1) We can use pure receive diversity to give us a better (less frequently faded) signal under conditions where there is a lot of fading (eg when the mean signal power is weak). This ALWAYS works.a2) We can use Alamouti or other space-time codes to make the above work even better, ie to give us an error-free signal even when the mean power is weak, IF the base station supports it.b) We can use MIMO (which again requires base station support) when the signal is strong, to boost our throughput.All the various 4G systems, like 802.11n provide for both a2 and b above. Only VZW seems to have the brains to use this technology on 3G systems (or at least to use it well — I know nothing about the Japanese system you describe).One final point. You may be interested to know that Apple's earliest 802.11n kit (base stations and computers) supported 2×2 MIMO. The most recent kit supports 3×3 MIMO. I am not sure on this point, but to the best of my knowledge they are leading (or were as of about two months ago) in this respect, with most (all?) PC and base stationvendors only supporting 2×2 MIMO. (This in spite of the multiple antennas sticking out of the base stations, which don't actually go to separate RF chains, and so are not independent and thus not actually useful.)

        • Niilolainen

          Appreciate you taking the time to write that.

          I understand diversity and fading etc, but the thing I didn't know was:

          "- Tests have shown that in urban environments, you get pretty much all the benefits of de-correlation at a substantial fraction of a wavelength separation. I forget the exact number — 15% or 25% — but it was remarkably small."

          I thought (based on the classic texts+ papers) you had to get up to a wavelength to get sufficient decorrelation, or start relying on crappy magnetic antennas etc. So that's real news to me.

          Thanks for the explanation of MIMO also. I was right– my understanding was very superficial!

    • unhinged

      I'm waiting for a "Mr Fusion" in 2015, as promised by Back to the Future… 🙂

    • berult

      The arrow of Time inexorably points towards the future. The future's arrow points towards the past, …where else? The more the future peers into the past, the faster Time eats into the future. The state of technology enables the present …as a future tempered by the past.

      Technology is literally the means to the end. Information horse-rides energy to primeval end-game perfection; any discernible pattern has to stand out against this overriding universal dynamics. If you encounter problems with energy conversion, storage and distribution within an information work flow, an organic masterpiece of a sort could save the day by deferring to its proof of content.

      Human cells, …we in a nutshell, quote from our genome how the future should reference the past …and whisper a definitive present to an infinitive collapse…   

      Transform, hoard, grid-lock, consume information and energy on the pattern of quintessential artistry; human cells networks tutoring technology on the fair usage of communication within frugal helpings of mitochondrial energy.

      If you buy into cellular, do it for what it's worth, …backward …Cretaceous compatible symmetry!