In my article "Intel Inside The Next iPhone? Not Likely, But Also Not Important" penned earlier this month, I countered the speculation that Intel (INTC) may become Apple's (AAPL) processor supplier for future generations of iPhone products. Lo and behold, the iPhone 5 was not Intel-based, but in fact powered by a custom Apple-designed CPU built at Samsung's fabrication plants. Not only is Intel not inside of the iPhone 5, but the likelihood of Intel (or any other chip designer for that matter) scoring the SoC design win in the iPhone has been reduced to what could only be a small rounding error away from zero.
This was, of course, expected, as moving iOS and all of the applications to be x86 would be a hassle and a flat out risky business decision. However, a much more interesting -- and important -- question is whether Apple will give Intel the boot when it comes to the MacBook line of products.
Why Apple Went Custom For iPhone 5
In order to answer that question, it behooves the savvy investor to understand just why Apple moved from an ARM-designed (ARMH) core to a custom one.
First and foremost, ARM is not particularly fast at releasing new designs. The Cortex A9 design was released in 2007, and the Cortex A15 design was available in 2010. It takes about 2-3 years for system-on-chip designs to appear in production from the release of the core designs, so Apple would be forced to continue to recycle existing technology for a couple of iterations before moving to a fundamentally new architecture.
Next, ARM does not have the resources to design specialized cores for particular types of device, so it produces "one-size-fits-all" low power micro-architectures. While SoC vendors using ARM's off-the-shelf designs such as Nvidia (NVDA) can tweak clock speeds and even make modifications to the design, it makes more sense for a cash-rich device manufacturer like Apple to simply design the best CPU for the device that it is trying to build by making the appropriate design choices and trade-offs.
Finally, as competition in the smartphone and tablet spaces heats up, there will naturally be some erosion on the margins of these mobile devices. An internal design will likely decrease the royalties that Apple has to pay to ARM which should lead to an increase in Apple's margins.
So, How About The MacBook?
While the R&D cost of spending 4 years to develop a single design for a very specialized, high volume flagship product can be easily justified. Can the same be said about the MacBook?
Well, the first thing to note is that Intel's notebook CPUs are built on a manufacturing process that is several years ahead of the processes available at the leading foundries. Despite the claims from GlobalFoundries that it will "match" Intel at the 14nm process node, the reality of it is that GlobalFoundries has always published aggressive roadmaps that have very consistently not matched up to reality. Taiwan Semiconductor (TSM) makes no claims to match Intel here.
The next issue comes down to the actual design of hardware. Intel has much more experience designing high performance CPUs. It's one thing to develop a custom low power ARM cores whose job is to match yesterday's performance in very low power envelopes, but bringing tomorrow's performance in relatively low power envelopes is a totally different ball-game.
Further, unlike ARM core designs, Intel has a very aggressive roadmap with each year bringing either a die shrink or a brand new micro-architecture. Apple would have a much harder time "doing it better" than Intel on the CPU side.
However, a hole in this argument shows up when one considers that the idea of "good enough" computing is a very real phenomenon. Apple could very well beef up the SoC that it uses in the iPad by ramping up the clock speeds, adding more cores, etc. and using that to power the MacBook products, even at lowered performance relative to Intel chips.
So, for the purposes of constructing a "worst case" scenario for Intel, let's assume Apple switches its MacBook lineup to an internally designed ARM-compatible core.
Impact On Intel If Apple Were To Switch
In the last several quarters, Apple has sold about 4-5 million Macs. So let's assume the worst case of 5M Macs/Q. According to Intel's 2011 10-K, Apple is not a 10% or greater customer, so for the worst case scenario, we assume that Apple's just shy of 10% at $5B in revenues to Intel. $5B isn't trivial, but it won't take down Intel.
Things get a little uglier when we see that while the rest of the PC space is struggling to garner sales, Apple is busy stealing market share. As Apple's share grows relative to the entire PC space, Intel's dependence on Apple's revenue stream becomes a lot higher.
A further consideration is the cannibalization of Mac sales, and in particular the MacBook Air, by the iPad. This would be a de-facto switch to ARM-based processors for Apple in the mainstream, while the MacBook line would become as the Mac Pro line is today -- a niche, high margin, low volume product that Apple doesn't really seem to care too much about.
The moral of the story is that if Apple were to switch its MacBook lines to internally designed ARM processors, this would have a noticeably negative effect on Intel. But is it likely to happen? Unfortunately, only Apple's management knows, but it's not so farfetched an idea. Heck, Semiaccurate's Charlie Demerjian believes it's a "done deal" and only a question of time. But then again, there have been a countless number of reportedly "done deals" that turned out to be not so done!