Rumors of Custom Mac Chipsets Implausible, Reflect Losing Strategy

By

post-2422-image-4e92e5d7e55b84ba24a82318a0651151-jpg

Image via Hardware Zone
The hottest tip on the rumor wires right now is that Apple does have really interesting Mac hardware on the way, as I’ve been known to suggest on a few occasions. Even Apple, at its conference call last week, was willing to acknowledge that it had a “future product transition” coming this quarter.

But the rumor circulating through the Intertubes this week goes further. It claims that Apple intends to use non-Intel silicon on its upcoming Macs. Not for the CPU, which will remain Intel, but for the rest of the chipset. While this rumor has slightly more credibility than it would if Apple had not recently purchased PA Semi or if AMD and VIA weren’t pumping out chipsets like crazy. And as AppleInsider notes, such a move could help Apple to differentiate based on silicon. Everyone else is using Montevina, and Apple could have something unique. It sounds like good judgment.

Except it’s a waste of time and money. Worse, it’s a losing strategy. After all, Apple doesn’t need to differentiate on silicon. Industrial design and software is enough. To read why, click through.


Hardware is only as good as the software that’s running on it and the external case and interface that users interact with. Amazing silicon doesn’t do anything unless the platform built on top of it takes advantage of its capabilities. And the great news for Apple is that other PC-makers are all running Vista or XP, and Apple is the only company that offers a multi-touch trackpad. In the worst case, Apple’s computers are competitive for the best-designed on the market. At best, they’re ahead by an interstellar mile.

To a purist, a Mac that has the same hardware under the hood as a Dell is some kind of disappointment. But to the rest of the world, it’s the difference between a commodity PC and a Mac. The overall design and engineering of the Mac, not to mention the beauty of OS X, completely sets Apple apart, regardless of what’s going on beneath the surface. The experience is just so much better.

And that’s why I will be shocked if Apple uses a genuinely atypical chipset with its next generation of Macs. If the people using those Macs, myself included, can’t see the benefit of Apple using its own supporting chips, there’s no reason to use them — focus on getting the best value for the dollar and optimizing OS X performance on Intel’s standard hardware. Apple differentiates through industrial design and superior software. That’s true on the Mac, and it’s true on the iPhone and iPod, too. Sure, the iPhone has a powerful chip in it, but an equally powerful chip is in the BlackBerry Bold — no one’s going to mistake one for the other. The look of the iPhone sets it apart, but even more so, OS X Mobile is in a class all its own.

People don’t buy computer chips — they buy a great mobile or computing experience. And Apple will use the most stable, robust chipsets that already exist until and unless they come upon a software feature that won’t run well without something exotic on hand.

As it is, this rumor just doesn’t make sense. Don’t swallow it!

52 responses to “Rumors of Custom Mac Chipsets Implausible, Reflect Losing Strategy”

  1. xav says:

    hmm…
    One should investigate another way : If that (hypothetical) chipset move occurs, it could be a manner for Apple to prevent the building of efficient clones, as they already did by using EFI instead of BIOS some years ago, which significantly slowed down their arrival.

    What’s more, that particular hardware move is IMO a good thing. Apple has such a mature softwae platform they can afford a ‘pause’ in its evolution. It seems likely that part of the work Apple plans for Snow Leopard is the integration of the upcoming chipset.

  2. John Muir says:

    OS X – whether on the Mac or the touch platform – is indeed Apple’s crown and what sets them aside from the competition.

    But acknowledging this fact does not then logically preclude this rumour.

    They bought PA Semi for something. Grand Central and OpenCL are promising software hooks into *hardware* power. It really does seem like their strategy is not just to rest on their laurels and rely on the OS X v Windows v Smartphones gap.

    Apple have a history even under the return of Steve Jobs in customising their hardware. It’s not all software innovation and industrial design. I think this rumour is more likely to fit with Oppenheimer’s statement than pie in the sky tablet Macs this quarter!

    Certainly, with PA Semi in house, custom is making a comeback at Apple for the long term. We’ll just have to see what happens in the meanwhile.

  3. Bill Olson says:

    I agree. Apple is fighting against the Clone Wars (I’m actually not a Star Wars fan) by building in proprietary chip sets that clone makers would have to reverse engineer which could cost a bunch of money, be a lot of hassle, and take a lot of time and remove any financial incentive to build them. I think that, as much as anything, is why Apple is doing this.

    I believe they were planning on this from the start but just didn’t have the people in house until a few month ago to do this.

  4. Peruchito says:

    i agree with you BUT i think that your statement applies only for ‘now’. in the future, when others will try to match OSX and apple’s designs, when other companies simply copy and clone apple’s hard invested product, what will differentiate apple products?

  5. leigh says:

    I agree with Pete, custom silicon is not a differentiators. (save for really custom custom chips like the new proc in Air).

    Also as a defensive move against cloning… that’s what Lawyers are for.

    the one place where custom silicone could come in handy is in hardware portability, that is to have iPhone, Mac, and all other future OS X based products *cough* iTablet *cough* running the same chipset, would lower cost of hardware and software support.

  6. Snafu says:

    I don’t think Apple is powerful and fast enough to develop its own PC-grade chipsets in a competitive basis. Arguably, transitioning to Intel freed it from the pains and the costs of developing the ones for PPC Macs, which were rather problematic (G5’s Firewire ports being significatively slower than G4’s, G4’s not supporting USB 2.x HiSpeed, etc.).

    I could see them trying to do something like that for some device that uses x86 processors but is not a conventional PC product, such as the much commented about Mac-class OS X based tablet. But I suspect such a product will use an iPhone-class OS X and probably will be Arm-based. Custom Arm processors, chipsets and Systems on a Chip seem to fit PA Semi’s expertise better.

  7. Louis Wheeler says:

    Product Differentiation. How does Apple differentiate itself from Wintel? You said that it would be “by software, not hardware.” But is that true?

    Yes, it would be expensive for Apple to design and implement its own hardware. The question is “do the benefits out weigh the costs?” Some of the hardware changes coming may make this possible.

    Microsoft has had a history of retarding both hardware and software development. MS tried to make a PC into a commodity, while Apple has tried to make it unique. Intel has tried to push the hardware envelop, with EPIC, for instance, and gained no cooperation from MS.

    Apple is a much better partner for Intel than MS. Both are hungry for change. Apple has the software arena well in hand with Snow Leopard. Snow Leopard will streamline the Mac OS, thus, creating a mature Operating System that Apple can build on for decades.

    Hardware improvement is one way that Apple can pull away from the competition. It can do so by working with Intel to drop the unnecessary and antiquated x86 hardware overhead. About a quarter of the space on the processor chips are used up in compensating for this obsolete architecture. Of course, that would mean that Intel would be producing a chip line that does not work with Wintel. Are the costs of doing this low enough to pay for the rewards? I don’t know. But, I won’t discount the possibility.

    If Apple & Intel can produce computers that are four to eight times faster than Windows computers at a cost that is a tenth less then Apple has no real market competition. The OS wars are back on again, this time, in Apple’s favor.

    All this depends on the costs to implement this. Processor chips are getting small enough that you can create them with sections that can be turned off. That way one chip can be used in many markets, so the development cost can be spread wide across many systems. It can be quite cost effective.

    Many of the future improvements coming are hardware related. Apple is preparing for this eventuality while MS is stuck in System 7 development hell.

  8. JD says:

    This is just rumor. However, relying strictly on Intel is the same mentality that has given MS an OS monopoly. Your strategy gives Intel the monopoly, and all monopolies eventually stagnate. Fortunately, Apple has strategies that allow creative people to innovate. Apple has been very successful at driving costs down on the iPod and iPhone by working with difference designers and manufacturers, driving the development of the best (cost & performance) components. Accepting the Intel reference package is the same thing the other PC manufacturers do, which is a huge part of the PC hardware. There’s almost no differentiation between Dell, HP, etc. Doing the same thing for the Mac will lead to Apple’s stagnation. Performing the same engineering for the Mac as Apple currently does for the iPod/iPhone should be expected – it’s in their DNA. But even beyond integrating the best chipsets (cost & performance), expect Apple to innovate new HW and SW combinations that will be unique. Otherwise, Apple is no different than a combination of a HP and MS – an unexciting future if so.

  9. Peruchito says:

    Louis is a smart guy.

  10. Snafu says:

    That would be G4/G5/Itanium all over again. What would be the point, for both companies?

    If Microsoft can be a harsh mistress (and at least it tried: EPIC simply didn’t match Intel’s claims, as some predicted well beforehand, so everyone left the room and Microsoft turned the light off), Apple absolutely can be an ungrateful dominatrix to Intel, as it was to Moto and IBM (and pay the price by getting just as much R&D from them as it deserved).

    Apple’s sales surged when Macs became Windows-compatible, the best of both worlds. So what now? Another transition? Another processor that ultimately won’t keep pace with the x86 world because demand won’t cover development and fabrication costs (however derivative such a “pure-Intel” chip would be, it still would need extra work done) ? And what for, exactly? A bit more speed? It can derive that from new GPGPU techniques already or Intel’s glitzy Larrabee thing next year. Whatever it needs for low-end hardware it can get it from other providers in an already RISC’ey fashion, such as Arm.

  11. James Katt says:

    I am all for Apple developing its own custom chipsets. I did this all along for years until it switched to Intel. Apple certainly has the money, the manpower (with PA semi engineers), and the knowledge to design its own chips.

    Apple can easily team with Intel to design the customized chips. It is Intel which may have problems since it may not have the flexibility to do custom chips for an individual customer. Intel – also – gave up it’s ARM business to Marvell. This is why Apple had to seek Samsumg out as a partner for doing ARM chips. Samsung had difficult with Apple’s specs since Apple made the mistake when transitioning to Intel of laying off its chip engineers. But now with PA Semi engineers in the fold, Apple has all the weapons at its disposal.

    Apple makes the entire widget. For it to follow like a lamb to Intel’s designs is simply not Apple.

    Moreover, Apple is primarily a hardware company. It is a fantastic software company. But hardware makes the profits and supports the software developments.

    For Apple to create a unique chipset and motherboard design is very much like Apple. Apple, moreover, wants THE BEST computer it can make. Intel’s designs are for general purpose use – not for optimal use – for every other company to copy. Apple wants to pull ahead of the pack. And Apple wants to pull way ahead of what cloners can do.

    It makes a lot of sense for Apple to do customized chips to improve its computational abilities while differentiating it from the pack.

    One idea I have is for Apple to do a hardware partner to Grand Central in Leopard. This would make it much easier and faster to use multiple CPUs.

  12. Neil Anderson says:

    PA Semi with a custom chipset to stop the cloners.

  13. Rufe says:

    Clone preventative, plain and simple.

  14. The Hague says:

    I’m thinking this is more about offloading certain processes, H.264, graphics, etc.

    Possibly part of the building blocks for QuickTime X.

  15. Peter says:

    Well, it obviously depends on what the new chipsets do.

    Consider, for example, a new high-speed/low power consumption GPU. Great for laptops, better than Intel’s X3100, maybe with it’s own high-speed memory. Suppose it adds an extra hour of battery life to a laptop. That seems like it would be an advantage for customers, right? 6 hours of battery life out of a 15″ MacBook Pro?

    So Apple should forgo that in order to maintain compatibility with Windows?

  16. Steve P says:

    Louis is a nut!
    4-8 times faster at a tenth less cost???

    He probably also buys those gadgets to help his car get 150 miles per gallon!
    There is such a thing as laws of physics, you know! (Or don’t you?)

    If Apple and Intel could do, in general, what Louis discussed – and work out something that would give Apple a consistent 10% edge on the competition – it would up Apple’s business imensely.
    (I’m not discounting the software tricks slated for 10.6, I just suspect that eventually MS and others will work out similar approaches. I believe Intel already has.)

    Still, there are certainly, now, a lot of good issues to speculate on. Fun! :)

  17. imajoebob says:

    Here’s a [wild] hypothetical for you:

    What if OS X 10.5.5 (or whatever ‘s called) is actually a test bed for OS 11? Or just that 10.5.5 is more than reported? And whatever iteration, it’s a true 64-bit system? Apple would have their own proprietary chips with the trademark co-development with OS whatever for the perfect fit. They won’t really compete with Intel, since they’ll likely only be used in Pros to start and still buy Intel for everything else. Intel will be selling boatloads of 32-bit stuff to PC makers who don’t need any 64s until MS comes out with Windows 64. In 2000 and twenty-what?

  18. iDave says:

    I’m with Louis wheeler: a hardware deferentiation would be a stake through the heart, if done correctly. Ever open up a Mac IIci? It’s a thing of beauty. Modular, snap-into-place design, proprietary everything, a modder’s dream. Imagine if you took everything Apple’s doing right these days and added customization and a little more thinking into the hardware aspect? Would Steve allow it? Probably not. But a little hardware fun can go a long way. The gloss on the apple’s outside (pun intended) isn’t everything.

  19. PIF says:

    PA Semi:
    As I recall, the DOD liked it for its very low power consumption, and, most of all, for its security features, unmatched by any other chip. And yes, it would stop cloners dead, and be far cheaper than firing up lawyers every time some cloner shows up (say the recent comments by MS’s Ballmer on managing the whole thing a la Apple).

    As one poster said elsewhere – evidently a developer – “You won’t believe it when you see it”. Which indicates to me that something else is also included.

    These features alone would fulfill the requirement that ‘…no competitor will be able to match’.

  20. Steve says:

    I’m no expert on these things, but I think people may be looking in the wrong direction. I was recently informed by someone way smarter about these things than I am that Apple will build into Snow Leopard GPGPU capabilities. I was also informed that GPUs, which are being used to build desktop supercomputers, can carry as many as 128 cores. Again, I am no expert. But I am informed, tangentially, that makes them orders of magnitude faster than conventional CPUs.

    So what if Snow Leopard takes advantage of a low wattage GPU for parallel processing along with an Intel CPU?

    So what if this new chip architecture, if that’s what it is, is just COTS GPU cleverly combined with a COTS CPU and Apple’s superbly tweaked Snow Leopard to create a blazingly fast, low power laptop? Or palmtop?

    Again, I have no idea about this kind of stuff, I just happened to talk to a couple of very smart electrical engineers on an entirely different topic. So maybe Louis or someone else who knows way more than I do can comment on the viability of this kind of approach.

  21. zahadum says:

    Pete Mortensen’s statement that hardware is not (going to become) a differentiator for apple’s product strategy has to rank up there amongst the stupidest statements on record (maybe on any topic, but certainly on technology).

    This kind of inane drivel is in the same league as Bill Gates (“who needs more than 640k”) or Mike Dell (‘fold up apple & return the money in the bank account to the shareholders’) or Mike Elgan (“apple and china are incompatible”).

    Not only does Mortensen fail to offer a /shred/ of evidence to prove his claim (‘custom chips implausible’) his analysis of Apple business model (‘a losing strategy’) is utter rubbish without any foundation at all ….

    The prosecution calls Steve Jobs to the witness stand!

    “Alan Kay had a great quote back in the ’70s, I think. He said, ‘People that love software want to build their own hardware.’ “

    – Steve Jobs at D5 May 30 2007

    cf: http://d5.allthingsd.com/20070… …
    … this is walt mossberg’s famous joint interview with Jobs & Gates (which EVERY real tech journo knows by chapter & verse, right?!).

    It defies reason how anyone like Mortensen, who has almost no serious journalistic credentials for covering apple, can utter such a completely ignorant, stupid, uninformed, unenlightened, counter-factual, and plainly absurd statement, and yet at the same time dare to pontificate about the inner mysteries of macintosh theology.

    Mortensen should be banned for life – not just from /talking/ about the macintosh … but even from /owning/ a mac!

    He is a menace to others & an embarrassment to himself.

    ‘Rock :: Paper :: Scissors’ …. ’nuff said.

  22. Snafu says:

    The thing is: in which areas does acceleration justify adopting a more expensive processor (it would be more expensive, not benefitting from economies of scale), recompikung and reoptimizing the OS and apps, making all developers go through another transition, and leading us users to see our equipment devalued once again, new Macs losing Windows compatibility?

    Better go the GPGPU (OpenCL) route.

  23. Pete Mortensen says:

    Dear Zahadum,

    Beloved troll, it would be in your best interest to up your reading comprehension ability. Nothing I said in any way contradicts what Alan Kay said so many years ago. My point was about chips. At this point in history, differentiating on SILICON, not on how the chips interact, what they’re used for or how software takes advantage of it, is a losing strategy.

    One-off specialized function chips end up losing vendor support and get abandoned. And they cost way too much money. Apple’s competitive advantage comes through how they engineer their own platforms based on the technology that other people develop and test. Period. A Mac Pro is beautiful because Apple makes the interior so accessible and easy to play around with — not because of the specific hard disk controller chip or South Bridge used in its design.

    Apple wins through combining software, hardware, and industrial design — not by foolishly over-investing in any one of the three. “Hardware” as Alan Kay meant it speaks to the complete device, not the individual piece of silicon — otherwise he would never have been excited about using Motorola processors in the original Mac. Going alone on silicon is a bad strategy these days, as Apple proved when it left PowerPC. It’s what that silicon lets you to build in terms of a comprehensive software/hardware experience that matters.

  24. Andrew DK says:

    @ zahadum

    Settle the hell down, buddy.

    @ Steve

    My thoughts exactly.

  25. Pete Mortensen says:

    Gotta be iPhone- and iPod-related, given the heritage of the existing processor in use there. Blending PA Semi and ARM tech could be a delicious blend for the mobile platform of the future…

    Still, I don’t think it’s related — unless the original rumor was just THAT garbled.