November 2009
Mon Tue Wed Thu Fri Sat Sun
« Oct   Dec »

Month November 2009

The Pocket PC Paradox

Recent data from admob shows that Android OS share is highly dependent on a single vendor (HTC). ( page 7)

There is an implied message in Android advocacy that once more licensees join in, volumes will be distributed among a broad set of companies benefiting from a healthy rivalry. This broad licensing will be a boon for volumes and for user choice and the platform itself will prosper.

I’m going to argue here that there will be no healthy distribution of volumes or profits in the Android platform.

History shows that volumes for an openly licensed mobile platform concentrate in the hands of a single vendor. To illustrate, I attached a list of all the devices which shipped (to date) with a flavor of Windows Mobile. The list includes 911 different phone models (search on your own on under the pdamaster tab–search criteria are on the last page of the PDF).

The Windows Mobile platform “took off” and had hundreds of concurrent licensees in the middle of this decade. In spite of this wide adoption, one vendor captured 80% of the volumes for all instances of the platforms: HTC. Having started with the iPaq as a PDA and followed by being the first to market with a WinMo device, HTC dominated Windows Mobile volumes even when faced with literally hundreds of competitors. These competitors included all the big names: from PC vendors like HP, Dell and Toshiba, to cellular incumbents like Motorola, Sony Ericsson, Samsung, LG and every ODM and OEM in between. The reason HTC won is mundane: they have the experience and the distribution to make an impact and they iterate rapidly and are intimate with the details of making a tough-to-integrate product work. They basically worked their way up the value chain from a contract design shop to a branded top tier vendor.

But if you exclude HTC, dividing the number of units of WinMo shipping at any given time by the number of licensed products in the market at that time, one gets about 50,000 units/licensee. Think about that: except for HTC, an average licensee could hope to ship 50k units for any phone they engineer and market. And they also need to prepay Microsoft the minimum license fee of $10/unit (approx) for a minimum order of 50k units or half a million dollars.

This is actually a paradox. Put yourself in the shoes of a business manager at one of the potential licensees. Consider the situation where there are 200 competitors in a market all shipping undifferentiated products, and one competitor has 80% share. Would you want to be the 201st? Saying yes implies hoping to split the remaining 20% 200 ways or obtaining, on average, 0.1% share? But this is not a share of all phones, it’s a share of that platform, which itself is, at best 10% of all smartphones which make up maybe 20% of all phones. So you are committing a few million dollars to pursue (.2*.1*.001=) .00002 of a market. That target is 3 orders of magnitude less than what Apple obtained in 2 years (without the margins). It never adds up unless you expect your entry to blow out of the 20% ghetto and take on the incumbents.

As it turned out there were dozens upon dozens of business managers who opted to be the 201st (or 301st). Unsurprisingly, the result was that every Windows Mobile business plan failed (except that of HTC). Every decision to join this ecosystem on the device side was a value destroying decision, repeated 900 times.


I don’t have an answer as to why so many are taken in by this. I have a hypothesis that it has to do with excess R&D capacity (e.g. idle engineers) that compels small companies to build product with no hope of market penetration, but I cannot prove it. They may also couple the low barrier to entry to the premise from the platform vendor that a no-name licensee could be the next Dell/Acer/HTC if the platform “takes off”.

Regardless of the cause, the proliferation of WinMo SKUs did nothing for the success of the platform overall. For consumer products, it does not matter what jockeying for position happens within the value chain if the end user/developer experience is painful and uncompetitive.

Android will win or lose on the user/developer experience. It will not win or lose on the proliferation of licensees.

Stack ‘em high!

In comparing two platforms like iPhone and Android, the question of addressable market for developers must come first. Apple has the advantage of an early lead plus its iPod touch devices which add another 40% to volumes. As a result, Apple created a 57 million units base with 100% yearly growth. It’s not a stretch to assume that before the end of 2010, Apple will have 100 million units in the field with 150 million by 2011.

All told, current Android licensees offered up less than 4 million devices to developers to date. When accounting where future Android volumes will come from, the chances are that Nokia, Samsung and RIM will not deploy Android widely as they have their own platforms to protect. That leaves Motorola, Sony Ericsson, HTC and LG and the current Windows Mobile licensee base. Even assuming unprecedented growth (270% y/y–greater than iPhone ever achieved) from this less-than-stellar list of vendors, it’s hard to imagine more than 30 million Android installed base by end of 2011.

One Android device for every five iPhone OS devices does not compel ecosystem switching costs for developers. Fragmentation and low volumes coupled with a tenuous vendor story leads us to conclude that iPhone will still have the upper hand by 2011.

Android Business Fashion Model

Nice summary: Essentially, Android isn’t Google’s phone platform, it’s an open alternative for failing hardware makers to use in place of Symbian, Windows Mobile, and Linux to create the same type of convoluted, fractionalized, and poorly integrated products they’re already making. This is also why Symbian, Windows Mobile, Motorola, and Sony Ericsson are all failing commercially.

Google’s primary and most significant contribution won’t be any major innovation in the core Android platform but rather in its own bundled apps, where Google plans to earn its revenues from via, to put it bluntly, adware and spyware.

Profit King

Imagine that!’s-most-profitable-handset-maker/


…One of the main reasons why UIs are unequally inferior are not only the way you build apps (open vs. closed hw/sw system) and the SDK itself but also marginal to non-existing UI standards, no ready-made drag & drop UI items, variations in carrier- & device firmware, hard- & software input, screen sizes, international customizations, modded phones, rooted phones and last but not least completely different expectations among users and the linux’ish target group itself. in a nutshell: beautiful mess. obviously, all these reasons eat up a huge pile of time that one could better spend with improving UX and polishing the interface. those who started early with android development have learned and are still learning it the hard way, just like they did with win 3.1 back in the days.


This prediction says that WinMo and Symbian are out at Samsung. Not a big surprise there.

The real news is some new “proprietary” OS is to take a majority share of the portfolio moving Android out of the picture gradually after 2011.

You will infer from my previous comments that the reason any deep pocketed vendor would shun Android is the same as why they would shun WinMo: no “control” over the whole product and hence no differentiation vis-a-vis an integrated competitor.

The Android devotees would then have to depend on their platform reaching critical mass without Nokia, Samsung, Apple, Palm and RIM, all of which will be shipping integrated products. LG got bought off by Microsoft and they will wallow in the mire.

Where does that leave Android? Moto and SE? They barely make up 10% of all devices and negligible smartphone share today and both are borderline going concerns.

That just leaves HTC, which shipped 80% of all WinMo. My bet is that they will ship 80% of Android, but that still tops them out at 10% share WW.

Looking at the platform game top-down and bottom-up you get the same story.

One Hundred Million

On April 9th, 2007 Apple announced that the 100 millionth iPod had been sold, making the iPod the fastest selling music player in history. The first iPod was sold five and a half years earlier, in November 2001.

Three months after that 100 millionth iPod, Apple sold the first iPhone.

My prediction is that sometime next year Apple will announce the 100 millionth iPhone OS device sold, making the iPhone the fastest selling platform in history. The iPhone will have been on the market for three years.

As of now there are at least 57 million iPhone/iPod touch devices in the field. This season there are likely to be over 15 million more sold. The bar to clear is therefore 28 million next year which seems achievable by Q3.

It goes without saying that 100 million contiguously addressable devices makes it the largest such device platform, but it may make it so by an order of magnitude.


Insipid and irrational politics are symptomatic to a company trapped by rigidity of response to asymmetric attack (and you can quote me on that.)

Nokia, the world’s leader in total mobile phone sales seems to be having new problems deciding on its OS strategy. There were some murmurings about embracing Android last summer, but Nokia seems to have decided against it. Stefan Constantinescu had this to say about the results of Nokia World 2009:

“The [new] software, Maemo 5, is a pain in the ass for developers since Nokia has admitted in public that Maemo 6 will come out in a year and it will break compatibility due to a switch from the GNOME environment to Qt. The browser, built on top of Mozilla technology, the same code that powers Firefox, is a step away from WebKit, the browser engine that powers Safari in the iPhone, the browser in Symbian, the browser in Nokia’s dumbphone OS known as S40, the browser in Android and soon the browser in RIM BlackBerry devices. Why is Nokia supporting something contrary to what the industry has already accepted as best in class? What’s the strategy?”

Technical mistakes and internal politics can sink a company in an intense competitive environment. Mr. Constantinescu didn’t perceive a coherent strategy. There may not be one.

Open Always Wins

That may be. Open only wins when the underlying service is a commodity for which improvements (other than price) will not be valued.

In my way of speaking, openness is semantically equivalent to “modular” and that is in contrast to “integrated”. Integrated is the only way to develop systems when they are not good enough. Modular is the only way to cheapen systems when they are more than good enough. Figure out where the technology is on the “good enough” trajectory and pick the winners and losers by the level of integration. It’s that simple.

There were hundreds of companies backing Windows Mobile for half a decade. And it made all the promises that Android is making now (non-phone devices, millions of developers, source code, Intel backing, contractors ready to build to order.)

The Android fan counter to this is (I’m guessing) that Microsoft could not execute! Google hired swathes of WinMo people into their mobile efforts. It’s the same crowd. Is execution some sort of magic pixie dust only available to Google?

The Symmetry of Google and Microsoft in Mobile

The question of Android viability goes deeper than the app ecosystem. In fact the ecosystem is itself dependent on the network effect of the platform. That effect is weak because the platform is not “tight” and is prone to fragmentation and its value cannot be communicated to end users. That is due to the lack of integration and consistency of purpose.

Bottom line, the problem with Android was always that it was a reaction to Windows Mobile. It was symmetric in its approach to the market with the added value of being free. In that sense it’s very successful. It might even gain all the share that WinMo used to have (about 14% of smartphones at its peak). But it will never rise above the nicheness of WinMo.

I’m always amazed at the technocrati babbling classes’ inability to spot causes. They go on and on about execution and never understand that execution is a resource not a strategy. Resources are fungible.

I for one do not “hope that Android succeeds”. It’s a foregone conclusion that it will not. It’s not a matter of hoping.