Over in Seeking Alpha, a contributor under the handle TechResearch recently suggested that Apple (AAPL 0.11%) would do well to boot Intel (INTC +2.40%) out of its Mac family of products and, instead, use custom-designed chips in the Mac.
The contributor argues that by ditching Intel and rolling its own PC processors, Apple would be able to "further differentiate Macs, broaden their appeal and improve sales."
TechResearch also claims that Apple would be able to improve its margins by moving to its own chips rather than pay for Intel processors, providing a boost to Apple's bottom line.
I respectfully disagree with TechResearch. Here's why.
A questionable "differentiation" argument
TechResearch claims that Apple could add workload-specific accelerators to its chips, which he claims would "further improve performance for certain types of workloads." This, he notes, would "help in improving the Macs' competitive positioning against Windows-based PCs," allowing Apple to boost prices of its systems.
Although I don't fundamentally disagree with the premise that, in controlling the chip design, Apple could do as TechResearch suggests, but I think that such direct control is unnecessary.
Keep in mind that Intel doesn't design its chips in a vacuum; customer requirements directly factor into the products that Intel defines. Apple is easily one of Intel's highest-profile customers as it buys chips in significant volumes and it buys a relatively rich mix of PC processors relative to what other PC vendors do.
I believe that if Apple identified a workload that it really wanted to accelerate, Apple is an important enough customer that Intel would work with Apple to get acceleration of such a workload built into the chips that Apple buys.
Would it be worth the pain to migrate Mac OS to an alternative architecture?
Mac OS has been developed with X86 in mind for many years now, and there is a substantial amount of Mac-compatible software out there that runs on X86.
Apple would, of course, have no issue porting over its own first-party applications to a new instruction set architecture; I'd be surprised if Apple hasn't already done that.
Where things get tricky, as TechResearch pointed out, is in getting third-party software developers onboard. This wouldn't be an easy task, but Apple is Apple and if the company were dead set on getting such a transition done, I am sure that it would be able to do so -- as painful as it would probably be.
Would Apple be willing to build many different chips?
If Apple were to transition the Mac away from X86, then I believe it would need to do it top-to-bottom, from the lowly MacBook to the expensive Mac Pro. In order to do this, Apple would need to develop several different processor families targeting the performance and power requirements of each Mac line.
To put it into more concrete terms, Apple would need to put together chips suitable for each of the MacBook, the MacBook Air, the 15-inch MacBook Pro, and the Mac Pro. The 13-inch MacBook Pro could use the same chip that the Air uses (which is what Apple does now with Intel silicon), and the iMacs could use the same chips found in the 15-inch Retina MacBook Pro.
Again, Apple could do it, but then it all comes back to the key question, "Why bother?"
Why bother, indeed!
If Apple was convinced it could build processors that were significantly faster/better than what Intel was willing and/or able to deliver, then it might be worth the hassles of an architecture transition as well as the spending associated with putting together multiple chips for different Mac product families.
However, I believe that Intel's manufacturing advantage vis-a-vis the semiconductor foundries that Apple relies on for its A-series processors, coupled with Intel's significant experience in designing PC-grade processors and platforms, means Apple would at best be able to deliver similar chips to what Intel can deliver.
Mac volumes are low enough that I'm not convinced that the "all-in" costs of building Mac silicon would actually be lower than simply buying Intel chips. Further, Apple would throw away the Mac's ability to run both Mac OS as well as Windows, a feature that is valuable to a number of Mac users.
And, finally, if Apple were to fail in delivering PC chips that performed on-par with or better than Intel's offerings, Apple would wind up trying to sell customers slower systems, to boot. Hardly a compelling value proposition!
Frankly, such a shift just doesn't seem worth it. Apple, in my view, is better served in focusing its silicon teams on areas where it can deliver solutions that are clearly differentiated relative to merchant solutions: iPhone, iPad, and Apple Watch.
