It’s almost exactly a year since I last discussed the possibility of Apple ditching Intel in favor of Macs powered by Apple-designed CPUs. I argued then that it was a question of ‘when’ rather than ‘if,’ echoing a view earlier expressed by my colleague Chance.
Bloomberg yesterday suggested that the ‘when’ might be 2020. That might seem like an ambitious timescale, but I do firmly believe two things. One, Apple is already running ARM-based Mac prototypes internally. Two, if it doesn’t happen in 2020, it won’t be too long afterwards …
By designing its own chips for iOS devices, Apple gets two core benefits.
First, efficiency. By designing both hardware and software in tandem, Apple can get far more out of its components than rival manufacturers. It’s often noted that Apple puts less RAM into iPhones and iPads than rivals do into their Android phones and tablets, and yet the benchmarks show that Apple typically performs at least as well as its rival’s flagship devices, and often significantly better.
Second, freedom from constraints imposed by third-party chip companies. Most Android brands are limited by what Qualcomm chooses to offer by way of smartphone CPUs, and when it chooses to launch them. Apple, in contrast, can do its own thing.
Those same benefits would apply equally well to Macs if Apple designed its own laptop chips.
I mentioned three challenges last time:
- Software compatibility
- Windows compatibility
Making ARM-based chips as powerful as Intel ones is still a challenge, of course, but it’s not an insurmountable one.
We noted in our review of the iPhone 8, for example, that its Geekbench scores were equivalent to a MacBook Pro, a point backed-up by others. And before the inevitable suggestion that phone and laptop scores are not equivalent, Geekbench founder John Poole says that’s old news: these days, they are.
In fact, the main limitation of today’s A-series chips isn’t raw power, it’s heat dissipation. iPhones have the processing power, but they quickly have to throttle back because they don’t have the heatsink capabilities of a laptop. Put the same chip into a laptop casing and that problem goes away.
This is the biggest issue by far. Either every Mac app has to be rewritten for the new chip architecture, or the new chip has to offer Intel emulation – which in turn demands even more power.
But, as I noted last time, Apple has done this not just once, but twice. First in the 1994 switch from the Motorola 68000 architecture of the original Macintosh range to PowerPC, and again in the 2006 move from PPC to Intel. It eased the transitions with a 68000 emulator and Rosetta.
That’s not to say that such transitions aren’t painful; they are. Horribly so for a while. Those of us who’ve lived through both moves can still recall some of the frustrations to this day.
But in both cases, we got over it, and the end result was worth it. The same will be true again. A key reason many of us buy Apple kit is for the ecosystem: the way that everything (mostly!) works seamlessly together. Having Macs powered by Apple-designed chips will make for even slicker integration between devices.
The Mac is expected to hit a 10% share of the laptop market this year. The macOS market is large enough, and its demographic attractive enough, to ensure that developers will produce apps for it.
The final issue is Windows compatibility. Given that Windows can run on ARM chips, albeit in a limited fashion, that problem too isn’t necessarily insoluble.
But there’s a second possibility here: accept the hit of losing what is an increasingly niche market. Yes, there are those for whom Windows is essential. There is specialist software that hasn’t ever made it to macOS, and which won’t make it to Apple’s ARM-based chip architecture. But that’s a small market.
Some might argue otherwise. In education, for example, there are still plenty of Windows-only apps. But as my colleague Bradley Chambers said to me this morning, education has increasingly moved to cloud-based apps. That process was greatly accelerated by the discontinuation of Windows XP, so these days it’s a relatively small number of very old apps.
And for those who need Office, Microsoft offered Mac versions right from the first Macintosh: Word in 1984, Excel in 1985 and PowerPoint in 1987. The Mac is a bigger market today, so there’s no reason to suppose that Microsoft wouldn’t support Apple’s new chipset – especially given how much of the Office suite is itself now cloud-based.
So I don’t think Windows compatibility is a show-stopper here, much as I expect to hear from the exceptions in the comments.
By the early 2020s, if not 2020
Apple has consistently demonstrated itself willing to take short-term flak for long-term benefit. It did so with its two previous chip transitions. It did so when it dropped optical drives. Again when it dropped the headphone socket from the iPhone. Yet again when it dropped older ports in favor of USB-C.
Apple doesn’t lack the courage to make big changes when it thinks they are justified by the long-term benefit. Dropping Intel in favor of its own chipset will be no different. It may or may not happen in 2020, but it is coming. If not in 2020, then certainly by the early 2020s.
There will be much anguished wailing when it happens. Lots of people (myself included, I’m sure) will be cursing Apple at various points in the transition. But the move is a logical one, and one that Apple will inevitably make.
Main image: Daniel Brunsteiner