Apple, Intel, and PA Semiconductor
There's been a lot of speculation regarding Apple's recent $278 million dollar acquisition of PA Semiconductor.
Some believe that Apple did it simply to regain technical expertise in chip design for its its consumer electronics divisions (iPods, iPhones) and others simply for the companies' patent portfolio. A few misguided souls even believe that Apple did it in preparation for an eventual move to its own custom processor designs.
Personally, I think the primary reason lies elsewhere.
Note that I said "primary reason". The lack of experienced personal definitely hurt Apple during the iPhone design process, and if one can say anything about today's Apple, it's this: they learn from their mistakes. They don't want the same issues to hinder new iPods, iPhones, iPads, and whatever else it is they have up their sleeves.
And while the patent idea has merit, I think there would have to be something awfully special hidden away in that portfolio to warrant the price Apple paid for it. Still, it has value, and there doesn't have to be just ONE reason to buy a company.
Which brings us to the custom processor idea popularized by Cringely.
Bob believes that the Intel move has hurt Apple as much as it's helped, at least in terms of hardware and system architecture. Prior to the Intel switch, Apple touted the superiority of the PowerPC design as a major differentiator between it and all of those backwards-facing Windows-based PCs.
Today, other than some often stunning industrial design, there's little to differentiate Mac hardware from PC hardware. Everyone uses pretty much the same processors, graphics chips, controllers, and drives. It's all "Intel Inside".
Before the switch, the PowerPC chip locked OS X to the Mac, ensuring that, unlike Windows, pirates couldn't really steal OS X and run it on their own hardware.
Today, one can download a cracked version of OS X that runs, more or less, on any beige box PC.
Bob believes that those too things have hurt Apple. And I agree.
His conclusion, however, is that Apple acquired PA Semi to force pricing concessions out of Intel and, as a first step in a glorious return to Apple's own personal private processor platform.
Interesting concepts to be sure, but I think Bob's added 2 + 2 and gotten 4.5. Close, but no cigar.
Apple could use a processor switch to wring a concession or two out of Intel, though it's pretty clear that both sides would know up front that the threat is an empty one. Short term, Apple is too tied to Intel to risk much bad blood.
So let's consider what would be gained, and lost, from a processor switch. In Bob's view, it's all gain. A really cool future processor differentiates Apple's platform and protects OS X from poachers. It could also, through some mystical marvel of processor design, emulate the aging x86 processor at full speed simply by throwing many, many cores at the problem.
The first set of arguments is compelling. The later, less so. The Intel switch made the transition to Mac "safe" for many users, despite the fact that the platform already had Intel/Windows emulators.
So what was the difference between now and then? Speed. Emulation works, but it's never as fast as the real deal. And the concept of throwing multiple cores at a single processor thread to do emulation is intriguing... but probably impractical.
And let's not forget that Intel won't sit idly by while all of this is happening. Their research budget is massive, and even without much competition (AMD) they've shown they're committed to improving their designs in the hopes of selling even more processors for even more purposes.
Worse, the last migration took nearly a year before Apple was able to covert all of it's hardware (MacBooks, mini's, iMacs, Mac Pros) to Intel. And almost another year before major third-party developers like Adobe and Microsoft managed to catch up.
Two years is a long time in the computer business.
No, another processor switch isn't in the cards. At least, not in this decade. Apple's developer base would revolt. The user base, forced to upgrade yet again to a new platform, would revolt. And Apple's stockholders would go ballistic, convinced that Jobs is certifiably insane.
But... with all of that said, we're still left with two valuable concepts: differentiate Apple's platform, and protect OS X.
So if it's not a processor, then what?
Good question, and one I'll answer tomorrow.
Comments