Apple has become a formidable chip designer with its A series chips. And that's probably not good news for Intel, says a chip expert.
Brookwood said, in effect, that if you reduce the relationship to the fact that Apple must continue to pay an "Intel tax" for chip designs it doesn't control, the future is not very cheery.
Q: What is the significance of Apple's A6 chip?
Brookwood: The A6 and A6x are the first ones that are using a [processor] core Apple designed.
Should Intel be worried?
Brookwood: The bigger issue isn't whether [Intel] x86 is better than ARM or vice versa, it has to do with business models. When Apple buys an x86 processor from Intel -- and I assume they get really attractive pricing -- they are paying Intel for the intellectual property embedded in that x86 processor. A so-called Intel tax. And they're paying for Intel's manufacturing prowess, in terms of being able to spit out really good chips. And [Intel CEO Paul] Otellini typically talks about that as collecting rent on both the IP (intellectual property) and the manufacturing.
And when Apple does its own chip, what happens?
Brookwood: When they designed their own chip, they had to pay ARM for a license, which is multiple millions of dollars for the architecture*, and they probably pay them a pittance in terms of per unit royalties for the chips that go out. Because basically it's not ARM's IP in the [chip] implementation, it's Apple's IP. So, there's a much lower variable cost associated with an ARM-based chip compared with an x86-based chip.
On top of that, Apple gets to choose which graphics engine they want. Do they want to have the Intel HD graphics or do they want to continue with Imagination Technology's graphics that even Intel is using in its Clover Trail low-power chips? So, Apple gets to make those kinds of choices about the chip, which they don't' get to make with an off-the-shelf Intel chip.
Is Apple uniquely positioned to go off on their own?
Brookwood: I think in this ultimate battle -- in the tablet and smartphone area but also MacBooks -- it's not so much the power and the architecture, it's the flexibility that folks, who have the volume to justify custom designs, get by doing their own. And that's not a possibility in the x86 world. You take it the way Intel or AMD think you should have it. You have it their way. And in the ARM world it's: have it your way.
Now, not everybody can do that because not everybody has the kind of volume that Apple has: taking a chip and using it in tens of millions of iPhones, tablets, and so forth. So, Apple can clearly justify the up-front R&D costs to do that. And the more units they can spread it over, the better.
What's the tipping point?
Brookwood: If Apple thought that they could have adequate performance to support the demanding applications that their professional users have. It's not just iMacs and MacBook Pros, this [chip architecture] would have to scale all they way to Mac Professional workstation class. It would be cleaner if everything moved (to Apple's chip architecture).
Any other scenarios?
Brookwood: I have this vision that at some point in the Apple-Intel relationship, it will switch from Intel selling Apple standard products to Intel selling Apple custom products designed by Apple and manufactured by Intel.
*Apple has an ARM architectural license.