Intel isn't inside the Apple iPhone

Apple has said the iPhone is based on an Intel processor. But which one?
Written by John G. Spooner, Contributor

Correction:  (1/10/2007 @ 11:45 p.m.) Intel is inside the iPhone. But its not there in the way I had originally thought it was.

I wrote the following post this morning based on a Reuters report (now corrected) that stated the iPhone would use an Intel processor. As it turns out, after checking with my sources, the iPhone does not use an Intel processor. (Apple TV does.) However, I still expect Intel to provide something of the iPhone—namely flash memory, which it already supplies to Apple for its iPod via IM Flash Technologies, its joint venture with Micron Technology. Also, I believe LPIA (Low-Power Intel Architecture) is something to watch. Although I think it would have looked a lot better in the iPhone than in any UMPC I have seen thus far, except Intel's most recent prototype.

Here’s the original post, unedited:

There’s still plenty of mystery surrounding Apple’s new iPhone. Apple, which unveiled the new phone/music player/ultramobile computer on Tuesday, has said the iPhone is based on an Intel processor, according to Reuters. Naturally, the next question that follows is:  Which Intel processor?

Thus far, I don’t think that either Apple or Intel is prepared to answer that. But I’ll make a prediction:  The iPhone uses a new processor—likely a super-low power, single-core derivative of Merom, the basis for Intel’s Core 2 Duo for notebook chips—that consumes 2.5-watts of power. Intel has already pledged to offer such a 2.5-watt chip in 2007. So I’m just filling in the blanks. I’m guessing the chip will use P1265, a special version of Intel’s 65-nanometer manufacturing process, designed specifically to produce low-power chips by cutting down on current leakage. Intel unveiled P1265 in September 2005 and said it would hit volume production in 2007. Intel also showed off a low-power chip platform, dubbed McCaslin, at its Intel Developer Forum in September, 2006. For what it's worth, I don't think the iPhone will use an Intel XScale chip, in part because Intel sold its XScale application processor line to Marvell and in part because I don't believe Mac OS X can run on XScale. (Someone correct me if I am wrong.)

Given that I have no insider information and the fact that neither Intel, nor Apple will probably announce details on the iPhone’s processor any time soon, I could be completely wrong. The iPhone could use a standard ultra-low voltage notebook processor, which would be the equivalent of the 5-watt Pentium Ms used in some of the first ultramobile PCs. But, even at 5-watts, I believe it would be difficult to achieve the advertised five hours of talk time and 16 hours of music playback with such a small-sized package. The iPhone measures 2.4-inches wide, is 4.5-inches tall and .46-inches thick. That doesn’t leave space for a big battery. Thus it seems to follow that the iPhone will run this new, as yet unannounced, 2.5-watt processor from Intel. It may also use bits of McCaslin. Intel, for its part, showed a prototype UMPC based on McCaslin, at the Developer Forum, last September.

Combine the iPhone win with the work by Intel’s Systems Technology Lab on to create a Low Power Intel Architecture which focuses on reducing the power consumption of system-level components, including LCDs, Intel CEO Paul Otellini reaffirming Intel’s stated goal of delivering processors that use less than 1-watt of power this decade and its move in July 2006 to form a low-power processor group, 2007 may be the year where the chipmaker’s low-power efforts comes together.

I can’t think of a better showcase for Intel’s low-power chips than the iPhone, particularly when I look at the lackluster debut of the ultramobile PC. Intel has been backing the UMPC along with Microsoft. But why not have a hand in both the iPhone and the UMPC. Ultimately, the iPhone will be greater than the sum of its parts, regardless of who makes them. But it’s the duty of this blog to point out how some of those parts might add up.

Editorial standards