Mobile Apple: The Limits to Growth (Part 3)

Even the largest avalanche is triggered by small things.

Vernor Vinge

It is better to err on the side of daring than the side of caution.

Alvin Toffler

(here's Part 1 and Part 2 of this series)

Mac: A Recap

In the previous part of this series, I left with a chart of the Mac unit sales over the last few years, contrasting those units sold into the existing installed base as replacement Macs, and those representing new users adopting the Mac platform. I'll begin this post with a similar chart, this time spanning a few more years back, and smoothing out the quarterly data through the use of a trailing 4-quarters window to compute the values charted. Also, I've included a couple years worth of estimates into the future (period denoted in gray) after a reader's request. Check it out (click to enlarge).

It's clear from this chart that the slow 5-year life cycle of traditional PCs, combined with the practically unlimited potential for expansion into the billion-and-a-half units within the overall PC installed base, precludes us from modeling any significant slowdown in the rate of adoption for Macs within the next few years. Perhaps only after a significant slowdown in wider PC adoption (currently at over 100m new users per year), and when Mac adoption surpasses 50% share of those new-to-PCs users (currently around 11%), one might consider an inflection in the incremental adoption of Macs (explore the charts on the previous posts to contextualize these two opposing trends).

To clarify, that would not be the saturation point of the Mac installed base, the upper limit or "carrying capacity," where Mac adoption would stop or perhaps stagnate at a rate comparable to population growth, but the moment when new additions to the installed base begin to decline, i.e. the top of that bell-shaped curve, the mid-point of adoption in which the installed base merely has reached half (in the case of a logistic curve) or maybe even a third (1/e ≈ 37% in the case of the skewed Gompertz curve) of the eventual saturation limit. If we assume that for the overall PC market that inflection was signaled by the "double top" in adoption in 2008 and 2010, and depending on how quick the disruption of traditional PCs by tablets plays out, we might not get the "mid-point" Mac inflection until 4-5 years from now, with several more years, perhaps even decades of continued but slowing expansion after that (unless that old form factor gets radically disrupted in which case we'd get declines in the installed base, see the next section on iPod for an example of this). Given this long lifecycle of 5 years, the installed base can continue to expand for several years even under Y/Y sales declines.

Such a long-term extension of the model, with those conservative assumptions, results in a saturation level (less than 2% Y/Y unit sales growth) for the Mac installed base of about 217m users, or about 3.5 times its current size, by the end of 2019 (remember it keeps expanding from there although minimally). I don't expect the broader PC market to expand much more from the current levels (maybe add another 300-400m users), allowing the Mac to reach a worldwide installed base share of about 11-13%, quite an ordinary range when compared to the current worldwide shares in the teens from HP, Lenovo, and even Dell at 12%. I would hardly characterize that as an ambitious projection (it does imply some significant penetration into the enterprise market).

Beyond this rich, long-term history of Mac adoption, I'll now explore the incipient adoption dynamics for the new mobile computing devices (iPod, iPhone and iPad). To create a consistent presentation framework I'll use the same chart template as the one above.

iPod: Windows Trojan Horse, Mobile Trial Run, And Exemplary Adoption Cycle

The iPod sales data provides the most varied adoption cycle dynamics, from slow incipient growth to exponential acceleration, deceleration, then quick saturation and even "negative" adoption or defection (declines in the installed base). As a whole it presents an ideal case study into these mostly empirical patterns we are considering as examples of the generalized theory of diffusion leading to saturation in technology adoption. The typical bell curve shows up beautifully (except for a slightly deformed right side), and the sigmoid shape appears twice, as a first phase of adoption followed by a second phase of declines, making the cyclic nature of adoption and disruption of a particular tech innovation quite evident. Here's the chart (click to enlarge).

I love this chart. Notice first the vertical scale going more than twice as high as the Mac version, showing how the iPod was able to break free from the constraints of the Mac's measly installed base, with volumes obviously implying significant penetration of mainstream computing. It established a beachhead that, despite its niche application as a digital music player, gave Apple a critically needed, and even privileged, place in the minds of users with increasing mobile computing aspirations.

Given its single niche application within personal computing rather than a platform for any number of mobile computing solutions (although it did become a kind of platform for myriad accessories through the iPod dock connector, which the iPhone and iPad also inherited), it expectedly reached an inflection point in adoption by 2007, soon after the exponential growth of the 2 previous years. After that it continued expanding while the replacement phase kicked in, and by the end of 2009 reached saturation.

We might hypothesize that if there hadn't been an iPhone (and later iPad) displacing it, the light orange bars representing iPod replacements would have sustained that 55m yearly unit sales for a much longer time, the dipping red bars of defections would not be there, and the line charts of growth would be asymptotical to zero. But inevitably, along came the iPhone (and lesser feature phones before) to supplant its need.

The increasing iPhone adoption, and then the iPad, is the main culprit for those red defection bars. In the first fiscal quarter of 2009 (by the way the time scale along the horizontal axis aligns with fiscal periods with each fiscal year ending in September) the iPhone surpassed the iPod in new user adoption (yet not in sales volume) and from then on it was clear, given the iPhone momentum, that the iPod would see continued declining sales for several years, at least until a new market equilibrium was established for the few specific jobs that the three stand-alone music player form factors (shuffle, nano and classic) could retain.

The defections continue to grow for now, but the rate of growth is decelerating fast. I've projected another inflection point within the next year or two, in which the yearly defections begin to decline, aiming towards such an equilibrium a few more years into the future, and perhaps even positive (though minimal) growth once again. The justification for this lies within the true platform capabilities of the iPod touch, and its possible future repositioning as a VoIP communication device (as public WiFi access might allow), when the iPod touch adoption (projected to account for about 85% of all iPods sold by the end of FY2016) finally compensates for the trickling iPhone/iPad cannibalization of non-iOS iPods.

Under this 2-year lifecycle methodology, the installed base of iPods peaked at 109 million users between June and September of 2009, and by last quarter it had declined to about 87 million. I'm projecting a new equilibrium (<2% Y/Y unit sales declines) at about 43 million users by the end of FY2017, this level mostly based on the estimated 22 million units sold for iPod touch for two consecutive years, indicating the installed base of touches may have also reached saturation and might itself be suffering from iPad cannibalization (touch installed base of near 37m projected at equilibrium). A repositioning of the iPod touch (or a new brand of iPod) as a VoIP communication device making use of (future) free and ubiquitous WiFi might reignite an expansion well beyond these equilibrium projections.

Again, the iPod not only acted as a trojan horse into the non-Mac computers base, preparing and even subtly conditioning the previously non-Apple user base for the true post-pc mobile revolution sparked by the iPhone and iPad, but also acted as a trial run for Apple towards learning and successfully managing all the complex processes and moving parts needed not just to enter the mobile phone industry, but also bring along real computing to it. This includes some very valuable know-how, from operational aspects like managing the increasingly complex supply chain and ramping up and down beyond millions and millions of units, to much needed sales and distribution issues like dealing with new third-party retail channel relationships (and keeping their inventory in check), to strategic insights like, for example, extrapolating from the "platform" experiences on the iPod accessories the need to enable and cultivate the iOS developer community. Most insightfully for an analyst (and likely strategists within Apple as well), the dynamics of its adoption/disruption cycle constitutes a great study for the potential evolution of other product lines in the future as it exemplifies almost every possible aspect and outcome of the diffusion of innovations theory (except, so far, the possibility of total depletion of the installed base, complete obsolescence and discontinuation of the product line), and it will continue to provide insights in all of these matters.

iPhone: The Billion-User Installed Base?

The iPhone constitutes the single most influential product line for Apple in terms of revenues, gross margins, and of course profits. Wall Street analysts' 11% consensus EPS growth rate estimate for FY2013 necessarily implies almost saturation of iPhone sales by then, which in turn implies rapid deceleration of adoption in the lapse of a year or two, and the inflection point of fastest adoption most likely having occurred in the last quarter. This assumption, and its inevitable conclusion, is in stark conflict with recent data showing continued increasing iPhone adoption, even at an accelerating rate in the last year or two.

However, I have tried to accommodate a very conservative view in which the inflection point (maximum yearly adoption) does occur by the end of FY2013 (the accelerating Y/Y growth rate suggests otherwise). The shape of the resulting bell curve is a little odd, the bump seems a little compressed when compared to the thickness of its left tail, and the relative Y/Y growth chart line shows a rather abrupt and arbitrary peak for the last quarter. A more gradual plateau and slight deceleration in this line would be the more sensible projection, although sharp turns in the Y/Y growth rates have in fact occurred as seen, for example, in the first chart for Mac by the end of FY2008 under a recessionary macro-economic environment. So here's the iPhone chart including such a conservative projection for the next couple of years (click to enlarge).

Again, notice the right scale going more than twice as high as the iPod chart (which was more than twice as the Mac). And notice the top light orange bars are still increasing considerably by the end of 2013, despite the dark bars reaching the inflection point. The symmetrical (logistic) expectation would be to at least double the height once more before saturation ensues.

I'd like to be completely clear: the slowing growth pattern on the right highlighted in gray is in no means implied by the historical pattern to the left up to the last quarter reported (FQ1 2012). Since relative growth itself has been accelerating recently (more than exponential growth), there's no possible way to infer an inflection point yet, much less a saturation level (carrying capacity) for the iPhone. Nevertheless one might infer it, rather arbitrarily, from other external data points like the size of overall mobile phone install base, or the size and speed of adoption of the smartphone installed base and its penetration against the former, and finally a guess on the share of this market that Apple might eventually command upon saturation. The particular projection above is my attempt of being as conservative as possible while retaining a roughly bell-shaped curve for the dark orange bars (yearly new user adoption) as indicated by the diffusion theory.

Asymco's Horace Dediu has created some amazing log charts about the "race to a billion users" (see here, here and here). In those charts he approximates users by the cumulative units sold, disregarding any notion of replacement modeling, in contrast to the main idea for user base modeling in my analysis here. However, more recently he himself has noted that for the most recent quarter, replacements probably account for 63% of iPhone sales in the US. One billion actual iPhone users (or subs if you will) is certainly not out of the question at some point in time given the latest estimate of almost 6 billion worldwide mobile cellular subscribers as of 2011. Many believe all of those subs will eventually become smartphone users, and assigning a 20% share of smarthphones (thus all handsets from the previous premise) to Apple is a no-brainer. However, I have some reservations regarding the definition of "smartphone" in that scenario, and Apple's participation on all segments of the whole worldwide market if that scenario did in fact became reality. In my opinion there will always be some percentage of subs that don't require a smartphone, only voice. Also, as mentioned in the iPod section, some percentage of usage could migrate to VoIP thanks to free, ubiquitous, roaming WiFi access points, eventually resulting in cellular subs cancelations. Finally, I don't think Apple is either ready or eager to take on the cheap, and extremely cheap, pre-paid subs like those prevalent in many parts of the developing world, which currently account for the majority of the active subscriber base.

The projection shown above, extended far into the future, yields an iPhone installed base reaching saturation (<2% Y/Y unit sales growth) at about 630 million (320m annual iPhone units sold) by the beginning of FY2020, about 6 years after the inflection point in the end of FY2013, which is roughly symmetrical to the 6 years of increasing adoption from 2007 until that inflection. This, again, is the most conservative I'm willing to be given the trajectory up to now and what I expect about overall smartphone adoption. A more realistic conservativeness would put the saturation at 800-900m users perhaps happening a few years further into the next decade. Certainly something could come up to disrupt the iPhone trajectory, but if that were the case I would bet Apple itself would be participating in such a disruption, so trying to bake in any of that in the model becomes a futile and worthless guessing game of arbitrary what-ifs, and predicting disruption is pretty tough.

iPad: 2y vs. 5y Life Means You'll Buy 2 or 3 for Every PC (Duh Tim!)

There's very little history of iPad sales, as we're only now approaching the 2-year anniversary of the original launch. Therefore not only we do not have any replacement data to filter out yet, we don't have a clear trend of the Y/Y growth rates, given it can only be computed for the last 3 data points. Further, the first few points of Y/Y growth rate over a trailing 4-quarter window timeframe are distorted by non-comparable periods (the first quarter worth of sales is just one quarter, yet its trailing 4 quarters is compared to sales over the following 4 quarters even though it still represents sales over just one quarter).

The incipient, though fast, adoption makes any educated guess of the projection much more uncertain. As with the iPhone, in order to approximate Wall Street's ridiculous consensus EPS growth estimate for FY2013, I'll try to project an immediate slowdown (but without it looking too abrupt) in yearly new adoption of iPads, something that so far is not at all suggested by the straight up ramp in adoption over the few quarters of history we have up to now, but it wouldn't be completely unprecedented either (see above how iPod adoption slowed considerably right after the hyper-exponential growth up to FQ1 2006). Here's the chart with the very short history and such a conservative projection (click to enlarge).

Once again notice the right scale, now back to half of the iPhone range, and a bit higher than the iPod (which included iPod's maximum sales level, while this clearly would need to expand much further to fit the eventual maximum for iPad). Because of the quick, almost straight up initial ramp and the early inflection within 4 years after lanch, this shape suggests to me more of a Gompertz-like curve. Remember, for Gompertz the bell-shaped curve was skewed to the right (the right tail lasts more than twice as the left) resulting in the inflection occurring at only 37% of the final saturation level instead of at half-way there for the logistic.

Extending this very conservative projection far into the future results in user base saturation (<2% Y/Y unit sales growth) at about 400 million by FQ1 2021. At an annual unit sales volume of 200m, and traditional PCs selling at a 360m annual pace at equilibrium (1.8b user base upgrading every 5 years), for Tim Cook to be wrong (about tablets outselling PCs at some point) requires that iPad share of all tablets stays above 55%. Because this seems quite possible (see iPod market share), and because I highly doubt Tim would be wrong in his assertion, this tells me the projection above is too conservative and way more than 400m users will end up joining the iPad base.

iOS: Greater Than The Sum Of The Parts

As a final exercise (phew enough of this post!), here's a chart combining the iPod touch (estimated), iPhone and iPad adoption into the single iOS category. Doing this makes a lot of sense in order to estimate sales (and its growth limits) of apps, tunes, movies, iCloud related services, other software, and more, not only to make financial forecasts for Apple, but also to provide app developers and other content creators a ballpark idea about the size and growth potential of their Apple-based addressable market. So here's that chart (click to enlarge).

Notice the vertical scale goes about 60% higher than for iPhone, and it would need to double again to fit the whole diffusion cycle. It mostly resembles the iPhone chart, with the growth line charts looking a bit smoother. Extending the projection far into the long-term future, we get near saturation (<2% Y/Y unit sales growth) by the end of FY2020 at well over a billion iOS devices in use.

So, I've shared here some very conservative and tentative timeframes for all of Apple's product lines, with terminal rates of growth (<2%) not appearing until about the end of the decade 7-8 years from now, and at saturation levels multiple times as big as the current ones. But I can think of 2 alternative questions about all of this, and without any further discussion, here are their answers given the assumptions laid out above as well as several other assumptions like ASPs, GMs, OpEx, taxes and such, which are required to translate unit sales into EPS:

1. What about those 2013 consensus estimate growth rates? What would be a more realistic (yet still very conservative) growth for that year? A: 33% (EPS of $68 vs. $51 for FY12)

2. When might growth slow to the 11% predicted by the consensus, if not in 2013? A: Q3 (June) of FY2016 (EPS of $104.50 vs. $94 for ttm up to June 2015)

This concludes this series on the limits to Apple's growth. Hope it was informative. I'll leave you with one final "insightful growth quote" like those I've been throwing at the top of these posts, courtesy of Britney Spears:

I did not have implants, I just had a growth spurt.