Rumor is that Apple is shopping around for a new manufacturer for the SoCs that power its mobile devices. And that Intel is interested in the business. I think it makes a ton of sense.
First, this doesn’t mean, contrary to some rumors, that Apple is moving its mobile devices to Intel x86 like its computers. Maybe someday, but certainly not at first. This would be a case of Intel making the ARM-based chips that Apple designs, instead of Apple having Samsung or TSMC produce them.
Presumably, Intel has the capacity to do it. Intel has more fabs than anyone else. And the idea of Intel making ARM CPUs isn’t unprecedented. When Intel bought DEC’s chip business in the 1990s as part of a lawsuit settlement, one of the properties it acquired was DEC’s StrongARM CPU. The StrongARM became very popular in applications that needed a small, low-power CPU with reasonable performance. Intel revised the StrongARM design into its Xscale design. I still have a couple of Windows Mobile PDAs with Intel Xscale CPUs in them. Intel sold the Xscale and related IP to Marvell in 2006 so it could focus on the x86. But for several years, Intel was a leading vendor of ARM CPUs, as strange as that seems now. Computer history has some strange twists in it.
Intel would prefer to sell x86 CPUs. But Apple isn’t going to move its handhelds to x86 this year or next and has to buy its chips from somewhere, since it doesn’t have any chip manufacturing capacity of its own. Apple sold 47 million iPhones, 15 million iPads, and, as best I can tell, at least 40 million iPods in 2010. I suspect that iPod number is way low, but we’ll run with it. That’s at least 102 million chips to sell, based on last year’s numbers, and those figures are going nowhere but up in the near term. Nobody in their right mind would turn down that contract. They’re inexpensive chips, but it’ll still account for billions of dollars in revenue.
Intel does want to get an x86 CPU into the mobile space, but a manufacturing deal with Apple doesn’t necessarily prevent that. It gives Intel a revenue stream until they’re able to make that happen, and insurance in case those x86-based plans don’t pan out.
The move makes sense for Apple, too. Right now Apple is suing Samsung, and Samsung makes a lot of phones, so they’d rather not enrich a competitor. Samsung manufactures its chips using a 45 nm process. Intel’s top of the line process is 22 nm. The switch won’t happen overnight, but the odds strongly favor Intel being able to beat Samsung’s process technology at any later date. The smaller process technology means the chips will be smaller, faster, and less expensive. The cost advantage would improve Apple’s margins, allow them to lower prices, or a little of both. The speed increase would give them a competitive advantage that they could use as a selling point–why buy a 2-core, 2 GHz CPU from Samsung when you can buy a 2-core, 2.5 GHz Intel-made CPU from Apple? Intel could also provide flash memory to Apple, much like Samsung is now, so they could continue to get most of the chips for the devices from a single vendor.
Partnering up with Intel never occurred to me, and it’s really out-of-the-box thinking, but the more I think about it, the more it makes sense for both companies.
I don’t see Apple moving their mobile devices to x86. Not any time soon, at least. They’ve bought two different ARM design houses in recent years in order to build up a design team, and they intend to use that to get a competitive advantage. Apple has a history of changing CPU architectures when it makes sense, of course, but they’re not going to do it until Intel can match the performance-per-watt that Apple is getting with ARM-based designs and at a lower total cost.
Apple is just as likely to move Macintoshes to ARM as to move its portables to x86. Frankly, either move is as much posturing as possibility. Apple likes leverage and they’ll say anything in order to get it.