On Apple’s Vertical Silicon Strategy
An alternative title for this piece could be ‘Why Vertically Converged Mobile Platforms Have Won Over Modular’. Over the weekend Ben Thompson wrote an eloquent essay relating the iPhone’s success to disruption theory in mobile / consumer markets.
Mobile has always been evolving differently, and 3.5 years ago I wrote a post called The Genius in Apple’s Vertical Platform referencing how Apple’s vertical (non-modular) approach allows it to innovate faster (my emphasis added):
“Apple’s DNA in this area is untouchable, helping it to innovate at the confluence of software and hardware. There’s a reason why pinching and zooming on the iPad is snappier than anything people have ever seen, and it’s not entirely clear whether the software or hardware plays the larger part.”
A few days ago we finally saw real life numbers on just how much snappier iOS7 is than Android—Apple phones are 2.5x faster than the fastest Android phones. Think about this for a second—in an age of decreasing latency in a “mature” market, such an advantage is almost unbelievable. Imagine how many times you pinch, zoom and scroll on your smartphone daily. Everything is 2.5x faster with Apple.
Is this performance gap endemic of a consolidated “good enough” market? The answer is categorically ‘No’. Say what you will about Apple vs Android in price or market share optimizations for low-end markets, one thing is absolutely clear: Apple’s silicon strategy is a secret weapon in allowing true converged experiences, consumers will pay for this, and in 2013 iOS continues to dominate competing platforms in system efficiency.
Since I wrote that piece, every other major mobile company has switched over to a strategy of vertical integration. But often their strategies are viewed superficially at the “box” level, or up the stack at services. So Google bought Motorola and makes handsets, that’s a truly converged platform play right? And now that Nokia has been rolled up, this enables a much better product, right?
What is missed is that despite others vertically integrating, Apple still has no current peers underneath the hood, at the silicon level. Let’s step back to 2010 again to a different piece I wrote detailing Apple as a platform company across software and chips.
“So the move to vertically integrate chip development helps Apple erect barriers and become a dominant platform company. This spells a larger trend – it is no longer adequate to simply be a device or software company to succeed.”
Despite today’s trend to vertical integration, Apple stands alone in its ability to integrate at the system level. The truth is silicon is the alchemy which causes consumers to marvel when they hold or touch an iPhone.
This fact is not lost on Tim Cook, who at the operations level understands silicon better than nearly any business leader. In this week’s Business week interview, this was very apparent:
“Now, we’re well beyond just the surface level of design of hardware and software. We’re deep in the guts. This week you saw the A7. You saw our new M chip. Well, these are only possible because many years ago we elected to start building our own silicon team, and now we have many, many people designing silicon.
And you saw us go to 64-bit. Well, why are we able to do that first? It’s because we’re at that level of being vertical. Does anybody—do these other three companies have silicon expertise? You can answer that. Maybe they have something that I’m not aware of, but in terms of the depth of it …
So it will be interesting in the next round, the next wave, to see what happens there. What do people do? When we looked at it, we concluded we needed to do our own stuff because we were dreaming of products that couldn’t be done with silicon that you could go buy. So we designed our own and built an incredible team.”
So the question now is does Apple’s understanding of silicon continue to enable Tim and company to dream of a future that others can’t. Can Apple continue to distance itself from “junk products” going forward, and will silicon drive this?
The data is clear at iPhone launch that this focus on custom silicon is allowing Apple to lead on performance and power—the A7 benchmarks show this in spades, and I detailed at the software level exactly how Apple performs this magic with silicon and software on its A7.
But looking beyond that, one of the extremely interesting trends to watch is how Apple manages platform control and buyer / seller bargaining power when it makes decisions on levels of modularity. Whereas others conflate vertically integrating as a single designation, it’s really a gradient of decisions that in sum make up the platform.
The most revealing find came from the Chipworks teardown of the iPhone 5s showing that the M7 processor (which I wrote about before it was deconstructed here) is actually an SoC manufactured by NXP Semiconductor. This is funny because Tim Cook calls it “our new M chip”. So who’s chip is it?
The truth is, it’s a collaboration, something very difficult to pull off with a vendor. Apple wanted hardware / DSP / sensor assistance, didn’t want to design and fab the chip, and turned to NXP.
Tim Cook knows that buyer and seller bargaining power are largely driven by a company’s market share. If you are Apple you have massive influence with suppliers. And somewhat ironically, Apple’s in-house chip development team gives it even more leverage with suppliers—they know so much about silicon’s limits. So when Apple specs chips it understands whether certain things are feasible say 1, 3 or 5 years out. They know what to license, build, invest in, or buy.
Together, Apple’s supplier bargaining power and silicon knowledge, mean that Apple can push merchant vendors harder, obtaining semi-custom chips that no one else can get. Make no mistake, the M7—though manufactured by NXP—is not available as a standard product. Even if something similar is sold by NXP in the future, it won’t be optimized at the system level.
These points of collaboration started years ago and allow Apple to essentially treat silicon as a commodity they control, adjusting ‘make vs buy’ strategies years ahead of competitors, while wielding incredible power over suppliers. The results in a halo around platform evolution at the silicon level that is unmatchable.
It is simply not good enough to dress up hardware, chips and cameras together with services and call it state-of-the-art. Sure, cheap Android phones are much better than what was available years ago, but are they as good as consumers want them to be, or as good as the benchmark (Apple) that consumers touch, feel and aspire to have? The answer is unequivocally ‘No’, which is why mobile computing and desktop computing markets continue to diverge even as the world moves from PCs to mobile.
Apple’s incredible silicon advantage will continue to help it to stay agile and maneuver in the face of unrelenting competition in the mobile world. The quantifiable evidence is there—whether it’s the iPhone’s 9M opening weekend sales or the 2.5x better responsiveness when you touch your phone, the proof is undeniable that Apple’s magic is alive.
What a great post. Surprised the comments aren’t pouring in!
You’re so right about this – and what’s so exciting is that mobile doesn’t just mean cell phones (you know this but I think it’s worth highlighting). This should seemingly apply for any wearable computing, etc and keep Apple ahead of the competition as long as what matters is mobile.
Interestingly, this advantage becomes less useful to them on the AppleTV (or mythical iTV) front since performance per watt (arguably) matters less in the living room…
In fairness, the 2.5x number appears to be kind of a typo (more like 2.1x), but 2x+ is certainly still a difference.
M7 is intriguing, but A7 has the potential to bring home the bacon for 2-3 years all by itself. All it needs is a little restraint by the iOS 7 team (who’ve chosen eye candy over pure speed in a few instances) and some eager developers to show what it’s really made of.
To be fair, he was referring there to touch screen latency figures, not speed of the iPhone, or that is what the link refers to anyway (well, it refers to a daring fireball link that refers to the article measuring touch screen responsiveness). I think I can agree it could have been spelled out more strongly (that it is the touch screen speed, not the OS speed), but the figures are correct (well, given the tools the company that did the research has).
Nice post.
Maybe not related to Silicon specifically…but another byproduct of Apple’s vertical strategy is reflected in this statistic:
– About 98% of iPhones in the world have iOS 6 or 7, which has Passbook and cannot be deleted. That’s maybe 300 million copies of Passbook.
– Less than 10 million copies of Google Wallet have been downloaded (Google cannot force people to use it).
– Samsung might be able to mandate installs of their Samsun Wallet, but that’s still only a few 10s of millons.
So, with hardly any effort, Apple is winning the mobile wallet wars by a 30-to-1 margin. Whether Passbook is “better” or “worse” that Google Wallet is irrelevant.
Really interesting thoughts, especially considering the fingerprint sensor and where they could go with that in gen II, III and so on.
Thank you for this article. While I agree with you that Apple’s vertical, integrated, silicon-based strategy is a powerful one — and perhaps even the best one — I have some questions about some of your points:
“Sure, cheap Android phones are much better than what was available years ago, but are they as good as consumers want them to be, or as good as the benchmark (Apple) that consumers touch, feel and aspire to have? The answer is unequivocally ‘No’”
//Is it really a ‘No’, and if it is, is it really so ‘unequivocal’? Let’s think about this for a second. What standard are we using to gauge this ‘no’? You hint at something about divergence… or perhaps it’s the massive number of iphones that were sold last weekend?
A few days ago we finally saw real life numbers on just how much snappier iOS7 is than Android—Apple phones are 2.5x faster than the fastest Android phones.
//Whatever happened to the ‘good enough’ theory? The fact is, most users didn’t know or even care about this 2.1x responsiveness thing. It was never a primary selection criteria for any user — it was just something Apple users had always enjoyed, though until now they did so without knowing the specifics.
Hell, even if Apple were to decrease it by another 20 milliseconds — thereby becoming 3x faster than the S4 — would anyone care? Aren’t there diminishing marginal returns for this type of thing. Just as the Moto X gets along just fine with an inferior display, I don’t think Android users ever cared about the response time in milliseconds.
This goes back to my first question — can we really consider response time to be a factor of this unequivocal ‘No’ which you purport, when hitherto everyone was oblivious to it?
//The final question I have is a technical one, not directly related to your article — just as it’s easier to squeeze in more DPI in a smaller screen, perhaps it’s easier to increase responsiveness for a 4 inch screen vs a 5 inch one?
I agree with you that Steve did not adequately justify his “unequivocally ‘No’” statement: “arguably no” would have been a better statement IMO.
On your point about “good enough theory”, you might want to have a look at this article about how intangible factors affect the purchase decisions for consumer products such as smartphones: http://stratechery.com/2013/clayton-christensen-got-wrong/ (Steve actually provides a link above)
Hi Steve,
Nice article about Apple’s strategy!
But I was wondering about the 2.5x faster of iPhone. If I correctly understood it only takes in consideration the touch time. It’s basically the time necessary to trigger something. It doesn’t take into account the necessary time to process and respond, that in my opinion it’s the time that makes the difference.
Is it really worth to leave in front but not be fast enough to finish before the others? Allow me to make a ridiculous analogy with Usain Bolt. He is never the best at the start but he always finishes first. I’m not saying that Android’s best hardwares like HTC One or Galaxy S4 are faster than iPhone. I’m just saying that the comparision wasn’t so fair.
Once again, great article as usual!
Really excellent post! This clarifies Apple’s silicon strategy and consequently it’s vertical integration far better than anything else I’ve seen so far.
[…] On Apple’s Vertical Silicon Strategy | steve cheney […]
[…] blogger Steve Cheney wrote a great post recently about how Apple leads the pack in vertical […]
I think a big piece is that they also cut out the middle man, especially when they move to a foundry model. If they are good at what they do, they do not need to be as efficient as other silicon manufacturers, and, they can add special features optimized with their software stack. They have the volume to make this model work.
Having said that, I worked at COMPAQ 20+ years ago, and we made our own graphics chips and there came a point, where not selling to the outside market made our internal chips inferior and spelled doom for vertical integration. It ebbs and flows, like everything else.
Steve,
I just discovered your site and enjoy your posts. I would be interested on your thoughts of Apple developing a MacBook Air with a touch screen and an iOS emulator? I am thinking of this in regard to Delta’s decision to go with Surface. I think the Surface is a great piece of tech, but it is too heavy to be a great tablet and too limited to be a great laptop. An Air with a detachable screen that is an iPad and when attached to the keyboard is a MacBook. I think Apple needs to bring the soon to be 1 Million iOS apps to OSX.
Bill
Great strategies are built around controlling scarce resources. The scarcest resource in mobile is battery life. That’s the main reason AAPL wants to control silicon.
INTL/others are still busy chasing Moore’s law (declining cost/transistor), while AAPL has recognized that Koomey’s law (power/transistor halves approx every 1.6 years). The silicon that powered the original iPhone can now be powered by a single 200 mAh button battery. AAPL’s game is to maintain a certain threshold of computing power while constantly reducing power. Dedicated silicon affords custom DSP algorithms, hardcoded vector computation all of which increase power efficiency by 10-100x compared to general purpose microprocessor for certain tasks.
AAPL is simply skating to where the puck is headed.
Awesome websites
we came across a fascinating webpage that you simply can get pleasure from. Get a glance in order for you
Read through was appealing, stay in touch
make sure you visit the online websites we follow, which includes this a person, given it signifies our picks from your web
Cambridgeshire wedding ceremony photographer
an example of our website visitors a brief at the same time back suggested the subsequent web pages
Recommeneded website pages
Below are a few in the websites we advise for our prospects
Recommeneded online websites
Below are some of your sites we highly recommend for our people
I’m reading this article two years after it was published, so today is can be described as more than interesting — it was prescient. Apple continues to widen its advantage in silicon, and that advantage continues to yield dividends.
Thanks for reading!