Despite stock price corrections in both Apple (NASDAQ:AAPL) and ARM (NASDAQ:ARMH) post earnings, together they are leading what is perhaps the most important (and much overlooked) disruption in tech: a manufacturing paradigm shift away from commodity processor manufacturers such as Intel (NASDAQ:INTC), to a new model of custom systems on chip (SOCs) designed by device manufacturers such as Apple. This disruption will be a source of new markets and expanding revenue for both companies, making them good long-term investments.
The Overlooked Disruption
The disruptive character of Apple's design and production of custom ARM processors has been largely overlooked because it flew in the face of conventional wisdom: custom processors such as Apple's A-series SOCs were an aberration, not a trend. Eventually, economies of scale and superior process technology would allow Intel to assert dominance of commodity processors in mobile devices as it had in conventional PCs. In May 2013 Ashraf Eassa, wrote "ARM Holdings: the Real Reason That It Ends Badly," predicting the eventual Intel triumph and suggesting readers short ARM. I wrote a rebuttal article, "ARM vs. Intel: Why it Doesn't End Badly for Either" in which I pointed out that the most important ARM processor makers were the vertically integrated device makers, Apple and Samsung.
Apple and Samsung were examples of what I called a paradigm shift away from the commodity processor maker such as Intel, and Intel-like commodity ARM processor makers such as Texas Instruments (NASDAQ:TXN), Qualcomm (NASDAQ:QCOM) and Freescale (NYSE:FSL). I asserted that ARM's IP licensing model allowed mobile device makers to design custom Systems on Chip (SOCs) optimized for the particular maker's device. The system integrator role of the mobile device maker had naturally moved to the level of the SOC since so many functions besides processing had moved to the SOC.
Most importantly, companies such as Apple and Samsung could move beyond merely licensing ARM's processor designs to designing their own processor cores for their SOCs from the ground up. In this approach, only the Instruction Set Architecture (ISA) is licensed from ARM. Apple's 64 bit A7 (used in iPhone 5s and iPad Air) is an example of this.
The benefits to the mobile device makers of custom SOCs were mainly in the area of providing a more seamlessly integrated user experience in which Apple devices have always stood out. Intel advocates have not understood how unimportant raw processor speed has become in mobile devices compared to ease of use and hardware/software integration and reliability.
Taking Stock, with Benefit of Hindsight
A year ago, my writings on this subject met with considerable skepticism, which just goes to show that if you want to build audience share, challenging conventional wisdom probably isn't the way to do that. In retrospect, my words in "Intel vs. ARM" strike me as particularly prescient, except in one regard: I vastly overestimated Intel's ability to compete with ARM in mobile devices.
As I pointed out in "Intel's Mobile Money Losers," there appears to be a significant cost disadvantage for Intel's current generation of Haswell and Bay Trail processors that kept them from making inroads into the ARM ecosystem in 2013. I admit that writing in early 2013, it didn't yet occur to me that Intel would be significantly cost disadvantaged.
To be sure, there are those who argue that Intel's cost disadvantage, if it exists, is merely temporary, and will be remedied by Intel's next generation 14 nm processor. Intel will become the dominant mobile commodity processor maker, just as it became the dominant PC processor maker.
I can't disprove this proposition, but it seems to run contrary to too many observable trends. Between them, Apple and Samsung control 52.6% of the tablet market (in unit shipments) and 46.6% of the smartphone market as of calendar Q4 2013, according to IDC. Clearly they have become the driving forces in mobile devices, and I don't see either of them being particularly motivated to see Intel reassert its processor dominance.
I say this despite the fact that Samsung did market some Intel-based tablets and has shown less interest than Apple in producing devices using its own SOCs. Samsung's US Galaxy S4 uses a Qualcomm processor. Samsung is still an important ARM device maker, and foundry partner to Apple, fabricating the A7.
Furthermore, as a member of the Common Platform collaboration with IBM and Global Foundries, Samsung will benefit when the ARM ecosystem achieves a 14 nm FinFET process similar to Intel's. I know there's no telling when that will be, but it doesn't appear to be more than a year away. Apple is particularly motivated to get its foundry partner of choice into 14 nm production, and is probably devoting considerable resources to this end. Flush with cash, Apple is in a better position than anyone to make this happen.
My sense is that the process advantage on which Intel's hopes for cost competitiveness depend will be erased before Intel gets much traction in mobile. I have a growing, though not particularly well supported, suspicion that Intel's x86 architecture, even in its much enhanced x64 configuration, has simply become obsolete. It had to happen eventually. The lack of mobile competitiveness is just the first sign.
And this is where the opportunity really lies for the ARM ecosystem, not merely to be the universal mobile platform, but the universal computing platform for the 21st century. Apple's A7 points the way to a more efficient computing platform than Intel that can serve in all capacities, desktop, server, mobile,and Internet of Things. AMD is due to release its own 64 bit ARM processors intended for servers. In the server business, cost and energy efficiency are everything, so I expect ARM 64 bit server processors to capture a significant (~25%) share of the server processor market by the end of this year.
Where does the spread of ARM into traditional computing stop? I don't see that it does stop. The world needs more energy efficient computing, not merely for servers, but for all computing needs. Once again, I expect Apple to lead the way, eventually introducing ARM-based MacBooks and even iMacs. Because Apple has done such a great job with its developer tools, creating multiplatform apps has become almost one-click easy for developers. Apple will lead the expansion of the ARM ecosystem into conventional computing because it can do so with the least disruption to its own customers.