Hypercritical
March 2013
Self-Reliance
The mobile market, everyone agrees, is the technology industry’s future. What’s not so clear is which company is best positioned to thrive in that future.
For smartphones in particular, the traditional metrics are confusing. Android has 70% market share, but Apple is taking 70% of the profit. Google, meanwhile, is not benefiting from Android’s market share dominance as much as Samsung, which recorded $4 billion in profit from its cellphone and telecom business in Q4 2012. In the same quarter, Google made less—$2.89 billion—from all its businesses combined. And when it comes to selling actual smartphones, the only two companies making any money are Apple and Samsung.
So who’s winning? When pondering this, I find myself thinking about dependencies. What is each company doing for itself, and in what ways does each company rely on others? I think this balance, much more than profits or market share, is what will determine long-term success. Let’s see how the players stack up.
Google: Mini-Microsoft
Google’s Android strategy looks a lot like Microsoft’s Windows strategy of yore—minus the part where you collect all the money. Google got the other parts right, though: create a viable platform, support it, evangelize it, and get as many other companies as possible to use it. That last part is made a lot easier when the OS is free and open source, of course.
In the PC’s heyday, Compaq, Dell, HP, Gateway, and others all killed each other selling PC hardware, grinding their profit margins down to almost nothing, leaving only a few players (barely) standing in the end. Microsoft, meanwhile, sat back and collected the same fat software margins from all of them (and from nearly all of their customers, as well).
With Android, Google seemed to posit that there was value inherent in being the platform “owner,” even if hardware makers didn’t pay for each copy of the OS. Android was filled with connections to Google’s (also free) services. More people using Android meant more people seeing Google ads, which meant more money for Google.
In the early days of Android, this theory looked promising. As in the PC era, hardware makers jockeyed for position in the nascent Android market. Individual fortunes rose and fell, but the number of Android activations just kept growing. So far, so good.
But unlike the early PC market, the Android market hasn’t produced a group of strong competitors duking it out at the top. As previously noted, only one company, Samsung, is making any money at all selling Android smartphones—and it’s making more from them than Google itself.
From the beginning, Google has shrewdly hedged its bets by fielding its own line of Android hardware. More recently, Google purchased Motorola, giving it its very own bona fide handset maker. Thus far, none of these efforts have produced Samsung-like numbers. But it’s clear that Google is unwilling to be entirely dependent on other companies to create the hardware that its mobile OS needs to be a complete product.
Samsung: Death From Below
Samsung seems like an Android success story. Previously better known in the US for its TVs than its smartphones, Samsung combined its hardware manufacturing prowess (and its shameless willingness to copy other companies’ design cues) with Google’s mobile OS to produce profitable phones that customers love.
Though the Galaxy line of devices would not be possible without Android, Samsung is far from Google’s ideal of a dutiful Android licensee, selflessly ferrying customers to Google’s services.
Just as PC makers used to insist on adding their own graphical shell or other brand-specific “enhancements” to their Windows PCs, most companies selling Android-based hardware products feel compelled to put their own stamp on the vanilla Android experience. Samsung is no different, steadily papering over the underlying Android OS with each new release of its TouchWiz user interface.
And why not? If Android is a money-loser for every other smartphone maker, Samsung is obviously doing something right. In its recent ill-conceived Galaxy S4 launch event, Android was barely mentioned at all. Samsung’s dependence on Android is clearly chafing.
Apple: Once Bitten, Twice Shy
In truth, Apple has been bitten more than once by its dependence on other companies. The viability of the Mac once depended on Microsoft’s willingness to produce a decent version of Office for it. Later, the Mac faltered multiple times when IBM and Motorola were unwilling or unable to produce competitive desktop and laptop CPUs. When Apple wanted to revamp its OS, Adobe and Microsoft were unwilling to port their software, forcing Apple back to the drawing board. Then there was that time when Apple asked another company to make a phone.
Like a lover who’s been betrayed one too many times, Apple has hardened its corporate heart against any form of true partnership. If it’s important, Apple wants to own and control it. When Apple does work with others, it insists on having the upper hand. iOS developers serve at the pleasure of Apple. Manufacturing partners must fight for the privilege of building Apple’s products, often using equipment Apple purchases for them. And, of course, Apple has its own mobile OS that runs exclusively on its own hardware. As God is its witness, Apple will never be hungry again!
Steve Jobs personified this attitude, which is why he felt so deeply betrayed when Google, his partner on stage during the iPhone introduction, remade Android in iOS’s image. After that, Apple’s reliance on Google for essential parts of its mobile experience simply could not stand.
The trouble is, online services have not historically been Apple’s strength. That’s why it partnered with Google, Yahoo, and others in the first place. It took Apple several years (and several acquisitions) to finally replace Google maps—and the results were not ideal.
There’s an old saying in business: don’t outsource your core competency. Or, as Joel Spolsky originally put it, “If it’s a core business function, do it yourself, no matter what.” This guideline makes it easy for a software developer to decide to outsource, say, catering and landscaping services. But what about Apple, with its historically well-founded paranoia about relying on outside companies for anything related to its actual products? What happens when everything starts to look like a “core business function?”
Sometimes You Can’t Make It on Your Own
Even among just these three companies, there are more than enough dependencies to go around. Google depends on other companies to make and sell the vast majority of the products that run its mobile OS. Samsung depends on Google to make and support the most important software component of its flagship mobile devices. Even the fiercely independent Apple still depends on Samsung to manufacture many of its mobile processors (for now…) and Google to provide web search services—and perhaps to give a little help with maps as well.
Back to the original question: who has the upper hand? Yes, there are dependencies in all directions—but not all dependencies are created equal.
Despite its recent success, Samsung remains in the weakest position. It clearly doesn’t want to remain beholden to Google, and that’s the right instinct. But I’m not confident in Samsung’s ability to completely divorce its mobile platform from Android. I just don’t think it has the experience or expertise to be a real platform owner.
Furthermore, while Android’s market share may be overwhelming, Samsung’s is not. Even if Samsung had the skills to take the reins of its software stack, it’d have to maintain compatibility with present and future versions of Android, lest it become just another low-volume also-ran smartphone platform.
Google’s present position looks weak, but it has two big trump cards. First, Google has proven to be one of the few companies capable of creating, popularizing, and supporting a platform. Despite all the skinning and branding by handset makers, Google is still the driving force behind Android. This power can only be negated by another company that’s willing and able to match Google’s Android efforts on all fronts: OS development, app store, developer tools, evangelism, the works. That’s a tall order.
Second, Google is still the king of online services. Apple, the biggest technology company in the world, just tried to replace maps, one of Google’s second-tier services, and barely avoided disaster. Microsoft, the former undisputed ruler of the tech sector, has been trying for years to challenge Google for the web-search crown, with little success. Maps and search are not obscure or obsolete services. If you can’t create equal or better alternatives—and so far, no one has—then you’re stuck relying on Google.
Google still needs hardware partners to maintain its Android empire, but we already have a model for how a software-focused platform owner can dominate a market. It’s harder to imagine a hardware maker dominating while relying on a software platform controlled by someone else.
Finally, there’s Apple, the jilted lover, feverishly working to eliminate any dependency that puts it at the mercy of a potential competitor. Apple remembers when Samsung was a great source of mobile CPUs and Google provided network services for iOS. Now look at those two traitors. No partnership is safe!
And so, in addition to developing its own OS, designing its own hardware, producing many of its most popular applications (built in its own IDE using its own compiler and language), Apple now has its own mapping service, is designing its own mobile CPUs, and is trying to get someone other than Samsung to manufacture them—all the while presumably eyeing its other parts suppliers and software partners warily.
Despite the bumps, Apple’s position remains strong. It’s got the best app ecosystem, competitive, trend-setting hardware, great adoption of each new version of its OS, and double the margins of the only other company making money selling smartphones. Oh yeah, and it dominates the tablet market too. There’s a lot for Apple to do in 2013, but at least it’s poised to succeed or fail on its own merits.
Looking out further than a year, the picture gets fuzzier. An unfortunate side effect of doing everything yourself is that every other company starts to look like an enemy. Realistically, Apple can’t do everything—or can’t do everything well, anyway. Online services are only going to become more important with time, so it’s understandable that Apple wants to be the master of its own destiny in this area. But it needs to improve much more quickly if it wants to even remain competitive, let alone catch up to Google. Failing that, it needs to find some partners that aren’t mortal enemies. (I’m sure Marissa Mayer would take Tim Cook’s call.)
In general, Apple needs to engage in more balanced partnerships that produce sustainable benefits on both sides. The switch to Intel CPUs is a good example, especially given how the situation has changed since the deal was first struck. In business, no strategic partnership is forever, but that’s no reason to avoid them entirely. And who knows? Perhaps Apple’s good relations with Intel will lead to its next great mobile SoC being manufactured at 22 or even 14nm.
Let’s just hope Tizen doesn’t come up during the meeting.
The Case for a True Mac Pro Successor
The xMac has been back in the news lately—the idea, if not necessarily the name. Whether it’s called a “Mac minitower" or a “Mac Pro mini,” we long-suffering Mac Pro fans are all looking forward to the “really great” thing Tim Cook told us to expect this year.
What almost no one expects is another straightforward revision of the existing Mac Pro, a gargantuan tower-style computer built with server-grade CPUs and RAM that pushes the limits of computing performance. Very few people want that kind of computer these days, and even fewer people actually need one.
On paper, the Mac Pro may no longer be a viable product, but it would be a mistake for Apple to abandon the concept that it embodies. Like the Power Mac before it, the Mac Pro was designed to be the most powerful personal computer Apple knows how to make. That goal should be maintained, even as the individual products that aim to achieve it evolve.
Why is this important? If Apple produces a new Mac that’s faster than any of its current models by leaps and bounds, will people suddenly buy it in huge numbers, choosing it over the laptops, tablets, and phones they prefer today? No. Is it because a very fast Mac can be sold for such a high price that its huge margins will make its profits significant, despite the expected low number of sales? No, that won’t happen either. Is a new, insanely fast Mac even guaranteed to make any money at all for Apple? Sadly, no.
So why bother creating a true Mac Pro successor at all? Good riddance, right?
Bean Counters and Car Guys
In the automobile industry, there’s what’s known as a “halo car.” Though you may not know the term, you surely know a few examples. The Corvette is GM’s halo car. Chrysler has the Viper.
The vast, vast majority of people who buy a Chrysler car get something other than a Viper. The same goes for GM buyers and the Corvette. These cars are expensive to develop and maintain. Due to the low sales volumes, most halo cars do not make money for car makers. When Chrysler was recovering from bankruptcy in 2010, it considered selling the Viper product line.
Why wouldn’t a company want to get a low-volume, money-losing product line off its books, bankruptcy or no bankruptcy? If you can’t think of a reason, you may be what is known in the auto industry as a “bean counter.” Luckily for Viper fans, Chrysler had a few car guys left. Here’s a passage from Car and Driver’s preview of the 2013 SRT Viper—the Viper that almost didn’t exist.
“I knew the very last thing Chrysler needed during our bankruptcy was a 600-hp sports car,” says Ralph Gilles, the 42-year-old president and CEO of SRT and senior V-P of Chrysler Product Design. “But I’m an optimist. I wanted to fight for a chance. We discussed it for a year. I got Sergio [Marchionne, Chrysler CEO] to drive one of the last Vipers. He jumped in and disappeared to God knows where. He came back 15 minutes later and said, ‘Ralph, that’s a lot of work.’ He meant it was a brutal car. But he didn’t say, ‘Good riddance,’ or anything. Then in late ’09, I showed him a video of a Viper breaking the Nürburgring record. He watched all of it and was impressed. I gave him a list of the supercars the Viper had put away.
The car guys won; Chrysler chose to keep the Viper.
Apple is not yet in bankruptcy, but every other reason that Chrysler should have run screaming from the Viper applies equally to the Mac Pro (except perhaps the lack of profitability; Apple doesn’t share that information about individual Mac lines). To understand Chrysler’s decision, let’s consider why halo cars exist at all.
One reason is prestige. Though few people can afford to buy a Viper, its mere existence makes the affordable cars from the same manufacturer that have even the mildest bit of sporting pretension slightly more attractive to buyers. Yes, this makes little logical sense, but it’s a very real phenomenon. (There’s a reason the term “halo effect” reportedly dates back to at least 1938.)
Halo cars also push car makers to their limits. Engineering teams must use all their powers and all their skills to create the very best car possible. This exercise inevitably leads to the exploration of new technologies. The failed experiments are forgotten, but the winners eventually find their way into more prosaic cars from the same manufacturer.
To Boldly Go
The Mac Pro is Apple’s halo car. It’s a chance for Apple to make the fastest, most powerful computer it can, besting its own past efforts and the efforts of its competitors, year after year. This is Apple’s space program, its moonshot. It’s a venue for new technologies to be explored.
Consider Larrabee, Intel’s project to create a massively multi-core x86-based GPU. Rumor has it that Apple was working on integrating the technology into a Mac Pro. Intel eventually scuttled the project, but consider what would have happened if it had taken off, reshaping the GPU market in the process. Apple would have had a head start on integrating the technology into its OS and application frameworks. Its drivers would have had their kinks worked out. When it became feasible to incorporate Larrabee technology into the rest of its product line, Apple would have been ready.
I intentionally chose a (rumored) failure as an example because that’s part of the point. Better to experiment on your niche product than your high-volume money-maker. There are plenty of success stories as well.
Think of all the technologies that debuted on Apple’s high-end Macs: hard drives, color, FireWire, multiple CPUs, multi-core CPUs, 64-bit CPUs, programmable GPUs, real-time video processing. All these features had a chance to get shaken out on machines that most people don’t buy. When they trickled down to “normal” Macs, Apple had enough experience under its belt to implement them competently.
As for prestige, perhaps you think the existence of the Mac Pro has precisely zero influence on the average MacBook buyer. The existence of the Corvette probably doesn’t affect the behavior of Chevy Malibu buyers either. But things change as you creep up the respective product lines, edging closer to the high end. The Titanium PowerBook G4 was all the more impressive for incorporating the CPU previously only available on Apple’s “supercomputer” Power Mac G4.
I used the present tense earlier when I said that the Mac Pro is Apple’s halo car, but that hasn’t actually been true for a while. By allowing the Mac Pro line to languish for so long, Apple has negated any possible prestige effect and abandoned an arena where it could safely push the limits of PC performance.
I know what you're thinking. That was then, this is now. The age of the high-end PC is over! But halo cars are even more absurd than high-end PCs. There are some pretty hard limits on car performance. Anything that carries a human around can only pull so many Gs before its fragile cargo gives up the ghost.
Compare this to computing power, which has no apparent useful limit. While car performance has increased by perhaps a factor of 5 in the past 50 years (and that's being generous), humanity has absorbed a million-fold increase in computing power during that same period without sating its appetite for more. (And that factor gets quite a bit larger if I add GPUs to the mix.) Computers are not “fast enough.” They weren’t when they were invented, nor when they got 10x faster, nor when they got 100,000x faster still. They never will be.
To be clear, absolute performance is not the only worthy technological frontier. Apple continues to push the limits on many other fronts: miniaturization, power efficiency, manufacturing processes, materials, and, of course, user experience. The same is true for car manufacturing, where fuel efficiency, safety, reliability, and even comfort are arguably more important axes of innovation than absolute performance (the limits of which can’t be legally explored on public roads anyway). And yet there they all are, those absurd halo cars, laughing in the face of logic.
Look Into Your Heart
This brings us to the final, and perhaps most important reason that halo cars exist, and that the Mac Pro—or its spiritual equivalent—should continue to exist. Let’s talk about the Lexus LFA, a halo car developed by Toyota over the course of ten years. (Lexus is Toyota’s luxury nameplate.) When the LFA was finally released in 2010, it sold for around $400,000. A year later, only 90 LFAs had been sold. At the end of 2012, production stopped, as planned, after 500 cars.
Those numbers should make any bean counter weak in the knees. The LFA is a failure in nearly every objective measure—including, I might add, absolute performance, where it’s only about mid-pack among modern supercars.
The explanation for the apparent insanity of this product is actually very simple. Akio Toyoda, the CEO of Toyota, loves fast cars. He fucking loves them! That’s it. That’s the big reason. It’s why the biggest car maker in the world spent ten long years and well over a billion dollars developing a car that almost no one will ever own—or even know about, for that matter. It explains why Toyota scrapped the LFA’s frame design and essentially started over with carbon fiber midway through the development process. (Talk about a Steve Jobs move.)
And perhaps it also explains why the famously cantankerous Jeremy Clarkson of Top Gear, a man who has driven nearly every supercar produced in the last several decades, recently called the LFA “the best car I’ve ever driven.”
I’m not here to convince you that the LFA is a good car, that you should trust Jeremy Clarkson’s opinions on cars (or anything, really), or that you should buy a Mac Pro. All the common reasons you’ve heard for Apple to abandon the market for high-end PCs are logically and financially sound. They also don’t matter.
Apple should keep pushing the limits of PC performance because it’s a company that loves personal computers. If Apple can’t get on board with that, then all the other completely valid, practical reasons to keep chasing those demons at the high end are irrelevant. The spiritual battle will have already been lost.
Fear of a WebKit Planet
I must confess, I was neither surprised nor disturbed by last month’s announcement that the Opera web browser was switching to the WebKit rendering engine. But perhaps I’m in the minority among geeks on this topic.
The anxiety about the possibility of a “WebKit monoculture” is based on past events that many of us remember all too well. Someday, starry-eyed young web developers may ask us, “You fought in the Web Standards Wars?” (Yes, I was once a Zeldi Knight, the same as your father.) In the end, we won.
As someone whose memory of perceived past technological betrayals and injustices is so keen that I still find myself unwilling to have a Microsoft game console in the house, my lack of anxiety about this move may seem incongruous, even hypocritical. I am open to the possibility that I’ll be proven wrong in time, but here’s how I see it today.
As much as I despised Internet Explorer for Windows, and what its simultaneous stagnation and dominance did to the web, I don’t think it’s the correct historical analog in this case. WebKit is not a web browser. It’s not even a product. It’s much more analogous to Linux, an open-source project that any company or individual is free to build on and enhance.
Linux, once a personal project created just for fun, now dominates the data center. It’s also in phones, tablets, game consoles, set-top boxes, and even (sometimes) PCs.
Is there a “Linux monoculture?” In some ways, yes. These days, it’s surprising if a startup creates a hardware product sophisticated enough to need an operating system and that operating system isn’t Linux. And let’s not forget that Linux has all but wiped out the proprietary Unix-based operating systems that once ruled the high-end.
Linux is the canonical open source success story. It succeeded for reasons that are now so boring they’re accepted as common sense. There’s still plenty of room for variation and innovation, but now all the significant achievements are shared with the world. If a company improves Linux, it’s not just improving its own products; it’s making Linux better for everyone. Linux let us “put all the wood behind one arrowhead” (to borrow one of Scott McNealy’s favorite sayings), but on a global—instead of merely a corporate—scale. (Funny how things turn out, eh, Scott?) Linux solved the Unix problem—for everyone.
WebKit fills a similar role. Thanks to WebKit, anyone who needs a world-class web rendering engine can get one—for free. And the products built with WebKit are as varied as those built with Linux. Even products in the same category vary wildly. Chrome and Safari, for example, have different features, different extension mechanisms, different JavaScript engines, different process models, and very different user interfaces. Opera adds yet more variation. And these are all just standalone web browsers. Consider all the embedded applications of WebKit, from game consoles to theme-park kiosks, and the idea of a homogenous, stagnating WebKit monoculture seems even more unlikely.
I haven’t forgotten the past. A single, crappy web browser coming to dominate the market would be just as terrible today as it was in the dark days of IE6. But WebKit is not a browser. Like Linux, it’s an enabling technology. Like Linux, it’s free, open-source, and therefore beyond the control of any single entity.
Web rendering engines are extremely complex. There are very few companies that have the expertise to create and maintain one on their own. (Again, the similarity to Linux is strong here.) I’m glad all those developers at Apple and Google are working on improving the same open-source web rendering engine, rather than dividing their efforts between two totally different, proprietary engines. Adding Opera’s developers can only make things better. The proliferation of WebKit will be a rising tide that lifts all boats.