"If we accept that Intel mobile processors were not technically disadvantaged, then there is only one way in which they could fail to be market competitive, and that's based on cost," Hibben writes. He goes on to cite a previous article of his in which he claims that "Intel's x86 architecture simply wasn't cost competitive with ARM (NASDAQ:ARMH)."
Here's why this explanation is, in this Fool's view, incorrect.
Understanding the anatomy of a modern system-on-chip
A modern system-on-a-chip targeted at smartphones is made up of many different intellectual property blocks, one of which is the CPU complex. To get a sense of what a typical mobile system-on-a-chip looks like, take a look at this shot of the Apple (NASDAQ:AAPL) A9 chip:
In yellow, I have highlighted the CPU complex of the A9 chip. Notice that, although it is sizable, it is hardly even the largest block on the chip (that honor belongs to the Graphics Processing Unit, or GPU).
If it is true that Intel's X86 architecture necessitates a die area penalty in the CPU core, in the context of a full system-on-a-chip that penalty just isn't going to be large enough to really move the needle in terms of cost. A competitive x86 product with CPU cores that are slightly larger than comparable ARM-based CPU cores could still do well in the market without being prohibitively expensive.
So, why did Intel fail?
The bottom line is that Intel's chip design teams did not produce competitive products. In some sense, the fact that Intel's products for mobility routinely saw severe delays (often by over a year) is arguably the biggest culprit here; a product that may have been competitive in year X is unlikely to be competitive in year X+1 in an industry where the competition brings out new products each-and-every year.
Beyond that, though, Intel's chips had competitive issues elsewhere. Intel never seemed to be able to successfully integrate an LTE modem with an applications processor, and to date Intel has not demonstrated an LTE modem implemented in its own manufacturing technology (Intel's modems thus far have all been built on third party chip manufacturing technology).
Taking it a step further, though, Intel never achieved leadership in many of the key intellectual properties required for successful smartphone chips (modem, image signal processor, low power graphics, and so on).
In short, a mobile system-on-a-chip requires success in a broad set of intellectual properties and not only could Intel not build those leadership intellectual properties, but it couldn't stitch them together into cohesive system-on-a-chip products for the smartphone market in a timely fashion.
Why write this?
If Hibben's claims are correct, that X86 has this fundamental and significant cost structure disadvantage relative to ARM, then this indeed has negative implications for Intel's core server, PC, and Internet of Things chip businesses.
However, I believe that this argument is flawed and one should not extrapolate from Intel's failure in the smartphone/mobile chip market potential failure in markets in which Intel has demonstrated considerable success.