Evidence is Mounting: Apple Will Convert the Macs to ARM CPUs

| Particle Debris

The time since most of the Macs have been updated can now be described as geologic. Is that because Apple doesn’t care about the Macs? Time Cook says “I Love the Mac.” Or, more likely, could we be in for another major architectural change? Evidence is mounting that Apple really will abandon Intel and take the Mac lineup to ARM. Here’s the assessment.

RIP Intel Inside

First, let’s look at the latest evidence from iDownloadBlog.

macOS Sierra code suggests Apple could replace Intel in Macs with custom ARM chips

That article nicely recounts the history of this notion, including this important section.

And in 2014, former head of the Macintosh division Jean-Louis Gassée hinted that the first ARM-based Macs could appear in 2017.

Gassée explained:

If we follow this line of reasoning, the advantages of ARM-based processors vs. x86 devices become even more compelling: lower cost, better power dissipation, natural integration with the rest of the machine. For years, Intel has argued that its superior semiconductor design and manufacturing technology would eventually overcome the complexity downsides of the x86 architecture.

But that ‘eventually’ is getting a bit stale. Other than a few showcase design wins that have never amounted to much in the real world, x86 devices continue to lose to ARM-derived SoC (System On a Chip) designs.

The Key Issues

In addition to what Jean-Louis Gassée explained abive, there are some other nuances to explore.

  1. Intel releases its new CPUs on its own timetable that doesn’t rigorously take into account Apple’s technical needs and product development cycle.
  2. All iOS devices use ARM. This change will make Xcode and development in general more coherent across platforms. It could lead to better synergies between iOS and macOS.
  3. The ARM processors in the latest iOS devices are 64-bit and more than competitive with CPUs from Intel in performance.
  4. Apple is the acknowledged expert in taking millions of customers though a major architecture change. The company did it first from Motorola 68K to PowerPC in 1998 and then PowerPC to Intel in 2005.
  5. The very long time since we’ve had updates to the Mac Pro, Mac mini, MacBook Air and MacBook Pro suggests (but doesn’t guarantee) that this architecture change has been in the works. Apple is taking its sweet time to do it right. That would include, perhaps, a Rosetta-like mechanism that allows current Intel-based Mac apps to run on the new Macs.
  6. Apple’s technical roadmap may well be diverging from Intel’s, one that has to take into account the needs of PC makers, not Apple. Perhaps Apple is done with this. Also, the nicety of installing Windows natively on a Mac with Boot Camp is a notion whose time has come and gone.
  7. Virtual Machine hardware in the ARM CPUs and an instruction translator would still allow Mac users to run VMs with Intel-based OSes like Windows and Linux. From what we’ve seen, modern ARM CPUs have the horsepower to do that.
2013 Mac Pro

A new ARM-based MacPro? Perhaps not the droid we were looking for. Image credit: Apple

That said, I strongly suspect that the MacBook Pros, rumored to be announced in October will be the last Macs with an Intel CPU, the Skylake series.

The question now is, will Apple simply release the new MacBook Pros without mentioning the transition? This is important because if Apple tips its hand at an October event, too many customers might decide to wait until 2017 to upgrade their MBP.

On the other, had, it could make sense to introduce an ARM-based Mac Pro or 5K iMac and suggest that the MacBook Pros will be the last Macs to make the transition in 2017/8. Mac customers would collectively breathe a sigh of relief to finally see what Apple is up to.

And if you need a new MBP, buy it now (in October). It will be viable for years. Meanwhile, Apple will deliver a multi-CPU workhorse ARM-based Mac Pro/iMac for developers to get the ball rolling.

This is a delicate maneuver for Apple because customers are hungry for new Macs, and the company would have to both plan for and announce these new Macs in the right sequence with the right messaging.

Now we wait.

Next page: The Tech News Debris for the Week of September 26th. The new Apple Echo chamber.

36 Comments Add a comment

  1. I do believe your take on Macs with ARM processors is the future of Mac. Unlike what has been going on over the last few years with Intel; Apple’s incessant desire to control everything possible can be better realized in this way.

  2. I don’t think this will happen any time soon. ARM is great in single cores and multicores. It doesn’t have floating point, it doesn’t have multiple execution threads nor the thermal capacity to handle any of this.

    Benchmarks don’t
    tell the story. These are different tasks. Maybe someday, but not in the near future.

  3. Additionally, Bootcamp and virtualization are not ideas whose time has come and gone. That is naive. Virtualization will only become more critical in the future, not less. And you’re absolutely wrong if you think Apple will deliver a Mac Pro with ARM. The Mac Pro’s problems are with the nonsensical design and the garbage Video cards they come with.

  4. Apple might do this–in fact, I wouldn’t be entirely surprised if they do this–but it would be foolish of them, and their sales would plummet as soon as people realized that nothing actually worked on them any more. Most games for the Mac, for example, are not ports–they run in an environment derived from WINE or its relations, and depend upon an x86 base. ARM CPUs are not yet nearly fast enough to provide decent performance doing instruction translation from x86 to ARM so that these capabilities would remain.

    Nothing would make me more likely to abandon the Mac platform. I don’t think I’m alone.

  5. This will happen but we should expect a lengthy transition.

    MacBook and MacBook Air are obvious candidates for starting the transition. This is almost in the “low hanging fruit” category. No huge need for big compute capacity or demanding video – so this would be easy.

    The Mac Pro is at the other end of the spectrum. It needs lots of crunch an that might be hard to do with ARM, especially given the tasks that these machines often do. I’d expect this one to be last in the line.

    belmikey: you’re right, but many games spend most of their time in code provided by libraries or the kernel, and not in actual application logic. Apple did a study of this back at the time of PPC->x86 transition and found that many apps were about 10% local code and 90% system/kernel/etc. So this hit might not be as bad as you fear.

  6. Bregalad

    I believe Apple is certainly moving in the direction or A-series processors in Macs. They want to control the whole widget and likely see improved compatibility with iOS devices as more important than compatibility with Windows.

    At the same time Apple seems eager to make further inroads in enterprise environments and moving away from the processor architecture and instruction set used in the vast majority of corporate hardware could be a huge mistake. When Macs relied on emulation you hardly ever saw one in a business setting. Once virtualization was possible Macs flooded into fields where they’d never been seen before.

    Maybe Apple truly doesn’t care about keeping Macs in business. Maybe they expect sales of iOS devices will be able to pick up the slack. Judging by the falling sales figures for iPads, however, that probably isn’t a viable approach.

  7. Makes absolute zero sense for a number of reasons, seems nonsensical, user-hostile and would actually make the Mac’s less attractive, have less software, more expensive and would drive pro-users like me, over to Windows.

    However it will make them 0.3mm thinner because ARM series chips are slightly smaller than Intel ones, so I expect the transition to start in mid-October.

    In five new sparkly colours, including jet black.

  8. The problem with most of the comments to this story is that you’re looking at it from your point of view and not Apple’s. Many of you have said that it seems like Apple no longer cares that much about the Pro market and this fits right in with a move like this. Other than FCP and Logic, what pro apps are left from Apple? How much money does Apple actually make from the pro market as compared to the consumer side? The answer is…not much. If they can keep 80-90% of their customer base after switching to ARM and still keep their same margins while offering lower prices across their line, the loss of the pro market as compared to new potential consumers is nothing.

    Remember as well that with iOS Apple is beginning to make inroads to the Enterprise market which has the potential to be huge. Apple is better suited to take a chance like this than they were when they switched from 68xxx to PPC or PPC to Intel. They have a tremendous war chest that can easily weather any financial storm for 5-10 years. They have a strong Mac market that has grown quite a bit since 2005 and there’s still a quiet revolt over the changes made since the release of Windows 8 and 10. People are looking for something different but don’t want to pay the Mac tax. If Apple can lower their prices and still keep their margins this could be a win for nearly everybody.

    Microsoft and Adobe will release ARM versions of their software for the Mac (assuming it happens natch). They’d be crazy not to. Heck MS is already there with iOS and Android version of Office

  9. Evidence is mounting that Apple really will abandon Intel and take the Mac lineup to ARM.

    I see reasoned guesses, speculation and discussions portraying a parallel to Apple’s PPC to Intel transition but no actual evidence.

    First, there is no way an A10 can perform the sustained computing that is often done on a Mac. It is good for short ( few seconds ) bursts of intensive usage but more intensive tasks are beyond it right now. Sure, Apple might have longer-term plans to couple ARM and Macs but the A10 isn’t it.

    Yes, the future of the Mac Pro and anything “Pro” ( and I mean really Pro not just the labels Apple slaps on devices that are borderline to not pro ) is tentative based on extrapolation of current evidence ( i.e. almost three years since a Mac Pro rev and cancellation of their only monitor with suggestions to buy competitor’s monitors ), so if the Mac Pro has been canceled it makes it easier for Apple to wield some future ARM product. Of course a top of the line 27 inch iMac currently can out compute ( and for longer periods ) an A10 and Apple keeps the iMac mostly up-to-date ( so there is no indication of its eminent demise thus giving Apple fewer Macs to address with its ARM ).

    So, while Apple *may* take the Mac to ARM, the question is when. The answer is when ARM is ready technically plus the other details need to be sorted out( Mac users who like to run Windows in Boot Camp is one, when licensing deals allow it and more ).

  10. Apple’s A series processors are approaching the performance of that which is used in the MacBook. So a future MacBook could use ARM, as Apple’s advances in performance are exceeding Intel’s jumps. It will be a few generations and years before we could see a Apple ARM derived design in a iMac and even longer before in a MacPro. All that being the case, more and more people are not seeking faster performance from their computers than what exists now, so Apple A series could catch up to that level. The switch from Intel based macOS based to Apple A series based Macs could be helped along with a system similar to the Rosetta system used up to Snow Leopard.

  11. The switch from Intel based macOS based to Apple A series based Macs could be helped along with a system similar to the Rosetta system used up to Snow Leopard.

    It will probably easier than that and it’s easy to imagine such a translator won’t be required because Apple already provides several layers of abstraction for AppKit and Foundation frameworks. Plus iOS( ARM ), already uses AppKit/Foundation, so many apps will rebuild with minor Xcode changes unless the developer is writing low-level, non-standard framework code.

    Even though a Mac ARM strategy is certainly possible, it isn’t clear to me if there is strong motivation to move away Intel. For all their delays, Intel probably offers Apple low prices given the volume of chips they purchase, plus they want to incent Apple’s purchase other Intel chips. Macs are only a small part of Apple’s business, and assuming they plan to keep producing Macs ( else why invest the time/money to change them to ARM ), only Apple knows if Mac ARM is justified at a business ( accounting ) level. I hope Apple starts building more innovative products with features we want but realize a large corporate public company maximizing profits and minimizing expenses ( consolidating CPU platforms is one such strategy along with the other synergies it confers ) is constrained.

  12. Paul Goodwin

    Not sure moving to ARM processors in Macs would be any less of a mistake than it was for Apple to move to the PowerPC. In the beginning there were all sorts of charts showing how technically the PPC chip development would far outstrip the Intel ones in speed and functionality. It never happened, and eventually the market was industrial embedded processing and about 1% of the computers sold as the Mac’s market share dwindled because of compatibility issues with the x86 world and limited software choices. There were some great PPC Macs, but eventually it proved to be a bad technical move.

  13. Scott B in DC

    To the vast majority, nobody cares what the hardware looks like inside the box. Only the geeks care. Most users only care that their computers work, that I can run their software, and that they won’t get any viruses. Other then the kicks the only community that would care are those with heavy computing needs, like the creative community. However, their needs are being satisfied with multicore high performance graphics cards. If most of the graphics and floating point computations are being handed off to these graphics subsystems, and what difference does it make if an ARM chip has flooding point or not.

    BTW, this is another back to the future moment. In years gone by, in the time of storage to put terminals which is equivalent to today’s web browser, floating point processors for separate from the CPU. The microprocessor industry championed the integrated processor chip. Now it’s being broken up again. Everything old is now new!

  14. Wulfkind

    I think this line of thinking that Apple is eventually going to go ARM over Intel has further credence when one realizes that Intel announced this year they are getting out of the phone and tablet CPU business. Their ATOM line of CPU’s were cut down iCoresomething CPUs originally made for desktops and laptops and these CPU’s were being marketed and sold to device manufacturers from NAS designers to phone and table manufacturers. Particularly with the phone manufacturers Intel literally spent BILLIONS of dollars in rebates and cross-marketing in order to entice phone ODMs to start using Intel ATOM CPUs instead of ARM. It was a spectacular failure and Intel bowed out this year. Now Intel is trying to pitch what is left of the ATOM line ( they’ve killed that name BTW ) to ODM’s in the 2in1 Hybrid tablet/laptop space in the Windows world.

    So….just like the PowerPC alliance began to stagnate, particularly in the mobile space ( think Motorola PowerPC 4 ) and in the desktop space ( think IBM PowerPC 5 series CPUs ) Intel has stagnated and dropped out of the key mobile space where Apple is seeing the most innovation and revenue. Apple HATES not being seen as the most innovative in any aspect of their portfolio particularly when they feel they are being held back by a third party…..so….why not go “in-house” with their own home grown ARM based CPU?

    So….in my humble opinion…..you will see an ARM base Apple Air in the 2017-18 time frame, an entire mobile portfolio based on ARM in 2018-19 timeframe ( think Macbook, Air, iPad, iPod, Phone and Watch ) and in the 2019-2020 range you will see the first Apple desktop with an ARM cpu.

  15. Scott B wrote: … the only community that would care are those with heavy computing needs, like the creative community

    There are lots of folks who purchased a Mac because they can run Windows applications natively ( i.e. without a VM ). They are just normal users and not necessarily the creative community. Apple would potentially lose these customers by switching to ARM.

    John M. wrote: I’m thinking about 8 or 16 modified A10s (16 or 32 high performance ARM cores)

    Good point but this would require major software changes( and many developers might be either reluctant or not skilled enough to implement – ( it is difficult even with GCD – Grand Central Dispatch ) because most applications run on a single core ( main queue in GCD ) and use only one core. Even some of Apple’s apps are written for a single ( main queue with the UI ) core. I’m not saying this isn’t solvable but it would be a problem today and it would be a major effort. Moreover, a solution would probably require a major investment by Apple and it’s not clear they have the will to do this.

  16. JustCause

    John M. wrote: I’m thinking about 8 or 16 modified A10s (16 or 32 high performance ARM cores)

    Apple doesn’t need to switch to ARM to do this, they can simply add ARM chips in addition to the Intel chips. The software to support different processing units is already in the OS.

  17. archimedes

    “Apple is the acknowledged expert in taking millions of customers though a major architecture change.”

    Not again! 🙁

    As nice as ARM may be, it’s really convenient to be able to run Windows for games and apps that need it.

    One great thing about Windows is that customers haven’t had to go through these architectural switches, which are painful. As much as I prefer the Mac, I have to say that Windows definitely has better backward compatibility in terms of both software and hardware.

  18. Virtual Machine hardware in the ARM CPUs and an instruction translator would still allow Mac users to run VMs with Intel-based OSes like Windows and Linux. From what we’ve seen, modern ARM CPUs have the horsepower to do that.

    Well then, you should be able to provide a fine list of real world devices that folks here can buy to test this out for themselves. You know, a nice laptop or desktop.

  19. John M in #7 wrote:

    Virtual Machine hardware in the ARM CPUs and an instruction translator would still allow Mac users to run VMs with Intel-based OSes like Windows and Linux.

    The likelihood Apple has written ( if they didn’t start it years ago it won’t be ready anytime soon ) and will embed it in an ARM CPU is very unlikely, both technically and at a business level, IMO and there is certainly no evidence to support such an endeavor. Nice fantasy though.

  20. geneking7320
    How about the Macintosh retain Intel chips (until they are not profitable) and Apple introduces a new line of computers with ARM chips named after a different apple?

    Hmmm…I can imagine the commercial:
    Hi there, I’m a Granny Smith…and I’m a PC
    Or I’m a Honeycrisp…and I’m a PC
    I’m a Fuji…and I’m a PC.
    I’m a Haralson…and I’m a PC

    Not seeing a lot of other Apples that would sound as good.

    Hi there, I’m a Braeburn…and I’m a PC

  21. If they get an ARM processor to be anywhere close to the performance of a decent Intel processor (i.e. not the processors in Airs or the MacBook Retina), then I might care.

    If Apple moves to ARM without getting their ARM processor performance drastically better, it’s not a viable option for me.

  22. Old UNIX Guy

    It doesn’t matter what CPU Apple puts in future Macs if they don’t reverse the downhill slide they’re on with the quality of their software. If it’s not iOS or watchOS they clearly don’t care. Switching to ARM CPUs isn’t going to help that. Dumping Craig Federighi and bringing back Betrand Serlet might (along with letting Jony Ive be the CDO for stores, $10K watches, and $100K cars ONLY)…

  23. @Old UNIX Guy: Agreed.
    @Andrush: True but Apple is about high profit margins which means squeezing standard vendors ( Intel, TSMC, AMD etc. ) for major price concessions. To secure price concessions, Apple avoids vendors selling items they won’t discount because they aren’t made is mass quantities. Virtually all knowledgeable folks agree it’s technically possible but a lot of factors suggest it won’t happen anytime soon ( read prior posts for some ideas ). Brian

Add a Comment

Log in to comment (TMO, Twitter, Facebook) or Register for a TMO Account