For those uninitiated, an innovators dilemma is when a new company innovates in such a way as to profitably enter the least desirable part of a specific market.
Initially, the new innovation is not capable of use in the higher-end parts of a market because it is unrefined, untested or unpracticed (or any other problems which comes with newness). The players who dominate the overall market (the "incumbent") watches as the new comer take over their least profitable segment of the industry. They typically let the new comer have lowest market since it has the worst profit margins; and besides, getting out of the business will help their profit margins.
The theory goes that since the new innovator can continually improve the low-end product, it can, therefore, introduce it into ever higher-end markets; the innovator can continuously attack the incumbent's least favorable market, and thereby, following the same logic, the incumbent will always give it up. That is, they will always be able to move up-market because our basic business logic suggests that you should focus on their high-profit margin business and not fight it out for the low-end where profits are slim. (If it is still unclear, check out this video. Or watch Clay Christensen lecture on the theory. Or check out the Wikipedia article.)
The conclusion is that large incumbent's tend to ignore the exact people they ought not to ignore.
Inevitably, the new innovator takes over the entire market by beginning at the bottom. In fact, and according to author of the theory Clay Christensen, this is essentially what Intel (INTC) prevented in 1998 when Intel introduced the Celeron processor. As a budget processor, it was to compete against AMD's K6 processor (AMD) in the low-end market: the part of the market which emphasized MHz per dollar. That is, instead of giving up the low-profit margin market as some business logic might suggest, they went down market to stop AMD from competing.
Is this happening with ARM (ARMH) and Intel?
First, while ARM Holdings plc was founded in 1990 and its SEC filings go back to March 1999, ARM only entered the public spot light recently:
We can further hypothesize that the processor market is a spectrum from the lower end, embedded processors and chips, to high-end server processors. (Disclaimer: while I believe that is fair, I am no expert on processors.)
Therefore, we suggest that ARM's embedded architecture took over the low end of the market before Intel saw the market for what it was. In this case, the low end verse the high end is not a spectrum performance per dollar (as it was in the case of Intel verse AMD in the late '90s); but instead of lower power consumption verse high power consumption; or rather, a spectrum of processor MHz per Watt. On such a spectrum, ARM Holdings dominates the low-power processor market.
"So why is Intel so behind ARM in the mobile space? Simple: Intel screwed up. Over the last decade, Intel spent years focusing on performance over power management, desktop over mobile. They were so focused on winning the x86 Gigahertz war against the likes of AMD that they didn't see the rise of ARM and Apple's Post-PC world coming."
While I personally think the slogan "post-PC" is propaganda adopted by pro-Apple users (and I am biased since I write software on Windows PCs), he is exactly right on his first point. They didn't try to compete in the power management focused market - it didn't appear to present a problem.
(And as an aside, what is brilliant about the theory of disruption and innovation coming from the bottom of a market is that you frequently can define the bottom in different ways. Intel, in its war with AMD, defined the bottom of its market as low budget processors. But they failed to see how the power consumption low-end would become a top priority for their company; they failed to define the low-end of the market in other, different, ways.)
It should be noted that both companies approach the market in different ways from the outset. ARM Holdings relies on other manufactures to use ARM patents to manufacture their technologies. It doesn't manufacture its chips. Intel has its own technology and manufactures its own chips. While Intel dominates the x86 ecosystem (19% of its 2011 sales are to HP (HPQ) and 15% are to Dell (DELL); and Microsoft Window's (MSFT) dominates obviously), ARM's ecosystem exists in a much more diverse crowd of partners and licensees (see an article comparing strategies here). The idea is that the network of business relationships establishes barriers to entry on both sides: the x86 universe on the one hand, and the embedded universe on the other.
The article linked above notes how small ARM Holdings is compared to Intel. That comparison is misguided. For one thing, the relevance of ARM compared to Intel will not be gleaned from a revenues comparison since ARM doesn't manufacture its chips. If one could get an aggregate revenue figure from the sale of ARM architecture chips by its partners, then one would have a clearer picture of the relevance of ARM in comparison to Intel. We do know that ARM processor division shipped 7.9 billion units in 2011 and its unit shipments have grown at an annualized rate of 42% since 2009 (ARM Holdings 2011 20-F, p. 35).
For Intel to try to squash ARM's position in the low-power market, Intel introduced the Atom processor in late 2009. Also, company strategy now includes "energy-efficient performance" in their "three-pillars of computing." Intel notes on the first page of their 2011 10-K:
Our goal is to be the preeminent computing solutions company that powers the worldwide digital economy. We believe that the proliferation of the Internet and cloud computing has driven fundamental changes in the computing industry.
With their absence in mobile computing, to become preeminent in the worldwide digital economy means one thing: mass production of low-power consuming processors for the phone and tablet market. To wit, in the risks section in ARM Holdings 2011 20-F states:
Some semiconductor companies have developed their own proprietary architecture for specific markets or applications. These companies may reuse their proprietary architecture to penetrate markets where ARM is currently the architecture of choice, or where ARM may be used in the future, making it harder for ARM to penetrate in the future. For example, Intel Corporation has developed the X86 architecture for use in PCs and laptops. With mobile phones becoming smarter, Intel is trying to capture the high-end smartphone market with a family of chips based on the X86 architecture [ARM is referring, probably, to Intel's Atom processors]. This could limit ARM's market share in mobile phones and could prevent any further growth into mobile computing devices. (Emphasis added, pages 7 - 8)
And, further on in Intel's 2011 10-K we find:
In smartphones and tablets, we face established competitors such as QUALCOMM Incorporated, NVIDIA, and Texas Instruments Incorporated, which deliver SoC ["System-on-Chip"] solutions based on the ARM architecture and complementary wireless technologies, as well as companies that incorporate SoC solutions that they manufacture. (Emphasis added, page 6)
While the relevance of the two companies is hard to glean from their respective filings, Intel's operating cash-flow is monstrous compared to ARM Holdings. Therefore, and as many other article author's think, Intel may have the sheer financial power to force their way into the low-power consumption processing market:
With Intel R&D spending for 2011 at $8.4 billion and ARM Holdings at £165 million ($254 million) it seems like Intel could sneeze and destroy them. But this, too, is oversimplified and ignores the fractured R&D spending across ARM's partners.
Because of this incredibly interesting processor architecture battle, I cannot make a calculated bet on either company--but this could be a battle for the ages.