No, Future MacBooks Won’t Have ARM Processors

arm processors

According to a recent Nikkei report, industry sources say Apple is thinking about taking another step into the semiconductor industry. These sources claim Apple wants to cut its independence on Intel when it comes to notebook chips. Instead, Apple would manufacture the processors itself, based on ARM architecture. Personally, I think it’s a bunch of hogwash that Apple would switch to ARM processors for the MacBook.

arm processors
Is it possible that future MacBooks could use ARM processors instead of Intel CPUs? Doubtful at best.

What ARM Processors Would Mean for the MacBook

If Apple did decide to switch to ARM processors for the MacBook lineup, it would boost Cupertino’s ability to make the notebook computers thinner and lighter. The ARM architecture provides better power efficiency, better mobility, and equivalent processing power. Apple could also integrate touch, fingerprint, and display driver functions.

With all of that said, it would mean the MacBook wouldn’t run current versions of macOS. In order to make this happen, Cupertino would need to do one of three things: turn the MacBook into an oversized iPad, fork new versions of macOS for mobile and desktop use, or ditch the desktop market altogether.

An Oversized iPad

This might not be the least likely alternative, but it’s certainly close. We’ve already seen that the iPad doesn’t really take the place of a laptop computer. Apple knows this.

This would be the easiest path to take, though, if switching to an ARM processor for the MacBook. Since iOS is written for ARM-based processors already, it would be a relatively trivial task to get it running on a MacBook with that processor architecture.

The problem is, such a device probably wouldn’t do well sales-wise, unless massive changes were made to iOS. An ARM-based MacBook would need to be able to run professional-grade software, something not really possible now. It would also need more transparent and efficient access to the file system. In short, I don’t see it happening.

The Fork in the Road

Next, Apple might fork macOS to support two different processor architectures. That would mean a version of macOS for ARM, and one for Intel. I say this because the industry insiders claim the move to ARM would only be for MacBooks, not iMacs, the Mac mini, or the Mac Pro.

I don’t see this happening. Remember, Apple’s been down this road before with the transition from PowerPC to Intel. As soon as it could reasonably do so, Cupertino yanked the rug out from under PowerPC Macs, leaving them stymied in OS X Leopard 10.5.8.

Writing an operating system for a new processor architecture is not a trivial effort. Maintaining it equally between two architectures is even less so. I really can’t foresee Apple wanting or even needing to do this.

Ditch the Desktop

The final option is to ditch the desktop altogether. Apple’s not showing any signs of doing that, though. Cupertino just released a new iMac and iMac Pro, and is supposed to be giving us a new Mac Pro in 2018.

The desktop Mac market might not be Apple’s largest profit-maker, but it’s certainly healthy enough that Cupertino isn’t going to abandon it. The desktop used to be Apple’s bread and butter, but now it’s the jelly. Who wants their toast without jelly?

Final Thoughts

Business Insider and Nikkei are both convinced Cupertino’s going to switch to ARM processors for the MacBook. I’m not convinced. I’d need to see a plan for how Apple would handle having the desktop on Intel while the MacBook ran ARM.

Right now, the move just doesn’t make any sense to me, and I doubt that it really does to Apple.

8 thoughts on “No, Future MacBooks Won’t Have ARM Processors

  • As others have pointed out, you’ve glossed over a few points and neglected to mention that Apple has a long history of switching architectures quite successfully (MCXXX to PPC, PPC to Intel).

    I agree that Apple is unlikely to fork their development—but that is exactly why I think a switch to ARM is very likely. Apple is currently forked! They had to use ARM because Intel wasn’t there when the first iPhone was developed. In the meantime Apple has slowly morphed from a computer company to a phone company. Apple is currently developing for two platforms, I believe at the expense of the X86 based Mac OS. Now, while a convergence of iOS and Mac OS is ongoing, I doubt they will ever be the same from a UIX perspective (your first scenario is unrealistically apocalyptic) but it would be easier to carry over low-level optimizations if both platforms were identical from a hardware perspective.

    The ARM processor was also not originally for mobile devices. It ran PCs just like Intel’s X86 processors, it just also happened to be more efficient and easier to scale down to low-power devices. Why couldn’t desktops be run off of ARM processors? The new Apple TV is already faster than the current Mac mini. Also, there are already ARM servers and high-end mainframes from IBM and SUN still run RISC based processors similar to ARM. Remember Apple didn’t switch to X86 for the Xeon lineup. It switched because it couldn’t put a G5 in a laptop. Unlike depending on Intel, who’s lack of suitable updates really killed the Mac Pro, Apple could develop its own designs in house to meet its own power/design specifications. For a company like Apple that’s a huge win.

    I believe what is holding Apple back is: A. x86 compatibility, there are too many user dependent of VM and bootcamp now; B. whatever licensing deal they have with Intel which probably also includes thunderbolt which is mostly Intel tech. With MS moving towards ARM and Intel loosening TB licensing with TB 3.0, both of those issues could be mitigated in a generation or two.

  • An interesting piece and with helpful and insightful comments. Quite refreshing!

    A few thoughts from my some-years-ago experience, and more recent observations:
    – macOS and iOS do have a common code base, with obvious variations
    – the death of Rosetta (i.e. PowerPC support) was contractual, not technical
    – Apple almost certainly now has the skill and knowhow to build its own Rosetta2 for x86 emulation on Axx CPUs. Even if not, the current owner of that technology is IBM – with which Apple has quite a good relationship
    – building fat-binaries is trivial, if that’s what you want to do

    Jeff is right that pro/power users rely upon certain features that only an Intel-native system is able to provide, such as BootCamp/VMware/parallels and a whole bunch of drivers that could be more readily ported to to OS X (Intel) than to PPC.

    But he misses the significant point that these folks don’t buy MacBooks. The people who do give not a whit about x86 vs. ARM and care nothing for Windows compatibility, other than data formats for Office and others. They want great performance from the commonly-used apps and great battery life.

    Apple could run a test of this theory by building a new ARM-based MacBook and continuing to sell the existing x86 one. All the existing x86 apps would still run [probably at about the same speed, I suspect] but the BootCamp etc options would be totally off the table. My guess is that it would be a resounding success. App developers could easily build x86/ARM fat binaries as they did years ago and the popular apps would be very fast. Let customers vote with their wallets.

    A few years ago [in another life] I developed stuff that absolutely required Intel x86. So I used VMware (and occasionally a real, actual PC) and that worked well. Now in retirement I support a dozen or so Mac-using family/friends and none of them have a hard x86 requirement. Some with heavy Adobe CC usage but there’s nothing inherently x86 about that.

    The bottom line is that if Apple can keep increasing performance faster than Intel, as has recently been the case, then Axx CPUs in Macs are a foregone conclusion. Not for all Macs probably but the low-end ones might suddenly be faster than the higher-end ones. I am sure that the folks at Intel are trying to work the problem. Unfortunately, x86 is their survival and yet their millstone.

  • The problem, Jeff, is that you make an awful lot of assumptions about this “MacBook,” how it works, who it’s targeted to and how it will be marketed.

    You’re also grossly underestimating how far in advance Apple’s vision is for both iOS and macOS.

  • Here is how tough it was to compile a fat binary under nextstep, and you better believe they can put this feature back in Xcode quite easily…. You just click on the output machine architectures you want, and bam, you get a fat binary that will run on any of those machines with ZERO extra coding on the part of the developers…

    1. That’s for building an app, though. Forking an entire operating system isn’t so easy, I don’t think. Kernel-specific libraries would have to be built, architecture-specific drivers. Yes, iOS is Darwin, but I suspect that by now it’s about as similar to macOS by now as FreeBSD is to Darwin. Which is to say, not at all.

      1. Hardly. I work for a Linux vendor in release engineering. I can tell you personally that once you put together the building and packaging infrastructure to support multiple hardware architectures, it’s really straightforward to add another one. Yes, there are new drivers to support. Yes, there’s instruction-set stuff you have to do to the back end of the compilers. But our Linux OS distribution is very much the same beast on every platform. System 390, PowerPC 64, x86_64, ARM, etc. We add new architectures (ARM 64, IBM Power 64 big and little endian) and retire old architectures (DEC Alpha, Itanium) with regularity.

        Apple has a different GUI on the Mac versus iOS/tvOS/watchOS, but the underlying software platform is identical. It’s just a matter of what gets included/not included in the platform-specific builds.

  • I totally disagree. First, internally macOS is almost certainly running on ARM/apple processors. Furthermore, iOS is basically the same OS. Third, apple did this before moving from powerPC to intel. Forth, the reason they did this before is that this is basically NeXTSTEP 15.0, and NeXTSTEP since 3.2 was able to compile any app as a fat binary, for multiple architectures at once, or any one in particular. Apple certainly has this ability, and dumping the OS to a new processor isnt too much harder than flipping a checkbox.

    The real problem with macOS on non-intel chips are all the bootcamp/parallels/VMware ability for vertical apps. Something that apple has shown they really dont give a darn about, but that real businesses and professionals rely on. The other problem is while the apple processors could do a great job on laptops horsepower-wise, they are nowhere ready for heavy lifting pro desktop hardware. Then again, apple has with it’s actions shown it doesnt give a damn about professional desktops either…

    TLDR; apple has a higher chance of doing this than a lot of people suspect…obviously, IMO. YMMV.

  • Great points. This would also mark the end of booting into Windows on Macs. I also feel that the transition for all the software, especially the pro software, would take to long if at all. I can’t see Adobe supporting two different versions of macOS (in that second scenario). I think Apple has already committed to two types of computing experiences, mobile with a touch interface and a traditional desktop that is more robust.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.