Apple On Intel, Part II

Continued from 'Apple: Intel Inside?'….

But of course what everyone really wants to speculate about is if Apple's going to give Freescale and IBM the boot and use Intel-designed and/or fabricated (note the and/or, hold that thought) CPUs in its future Macs. Let's look first at whether or not Apple's going to embrace products in Intel's current CPU line, or next-generation CPU spins we already know about. Depending on which segment of the Mac product line you're talking about, this scenario could play out in different ways:
1) iBook and PowerBook laptops: The Freescale G4 CPU still hasn't hit 2 GHz, and it uses an archaic 167 MHz front-side bus. A dual-core G4 is on Freescale's product roadmap, and OS X and many of its apps are multi-thread cogniscent, but there's been no tangible sign that Apple plans to adopt the chip. There's been no sign, either, that Apple's been able to solve the "mother of all thermal challenges" (a mid-January comment by Tim Cook, Apple's VP of worldwide sales and operations) that hinder the company from shoehorning a G5 into the iBook and PowerBook form factors. Wouldn't a Pentium M processor, with its clearly demonstrated strengths in the MIPS-per-watt arena, nicely fit the bill? Yeah, I think so too.
2) G4-based eMacs and Mac minis: Here the primary motivation would be cost. Microsoft's Windows XP O/S (along with its predecessors) is widely viewed (by average folks, not just tech-heads) as a spyware- and virus-plagued piece of….well, you know. The Mac OS is viewed as spyware- and virus-immune (regardless of how much, or I'd claim, little reality is behind that belief). Steve Jobs knows that with this perception, coupled with the oft-touted iPod Halo effect, he's got a chance to make another run at mainstream success. The Mac mini is his first strike. But to keep going, he's got to keep driving prices down, and to keep profits up he's got to keep lowering costs in lock-step. Freescale-plus-IBM CPUs service less than 3% of the world's PC sales. Intel ships CPUs into over 80% of the world's PC sales. High volume = low cost. Celeron. End of story.
3) G5-based iMacs: These are the mid-range desktop systems. Performance is more critical here, but power draw (and corresponding heat dissipation) are also a concern given the current all-in-one design squeezed into a 'thick LCD' form factor. Unless the form factor changes, I view this as the segment of Apple's product line least likely to switch away from the G5 (or, said another way, the segment which'll switch last). Current-generation Intel CPUs, with the exception of the Pentium M, have their own well-documented heat issues. But dual-core combined with slower per-core clocking, along with architecture tweaks, will sooner or later get this issue under control. This is where I see Pentium D going.
4) G5 and dual-G5 Power Macs: On the one hand, a conversion to Intel seems bizarre here, given Apple marketing's rabid push of the G5's Velocity Engine and other claimed performance advantages over Intel CPUs, the assertions that the G5 Power Mac is a supercomputer for the desktop, etc. And from a recent slip-up in a revision of an Apple benchmarking tool called CHUD, we know that Apple's at least considering IBM's upcoming dual-core G5 chip. But two years removed from Steve Jobs' (subsequently, and frequently, repeated) claim at the 2003 WWDC that the PowerMac would hit 3 Ghz within a year, 3 GHz is nowhere in sight. That's embarrassing. Embarrassing enough to force a switch? Who knows. Most folks seem to think a high-end x86 chip, such as a Pentium Extreme Edition or a Xeon, would go here. I prefer to think different: Itanium 2. Three words why: floating point performance.

With all that said, here's one other possibility. Apple got burned big-time when then-Motorola badly stumbled on its product spec, cost and volume forecasts a few years back, a situation that ironically drove the company into the arms of IBM in the first place. What if Apple's contract with IBM gave Apple IP rights to the PowerPC architecture, and the ability to pursue alternate sourcing of chips based on that architecture if certain cost, volume and performance goals weren't met by IBM? With Intel ramping up 65 nm process technology as I type this, it's one of the only companies in the world that can stand toe-to-toe with IBM as a semiconductor manufacturing powerhouse (and, arguably, even beats IBM to the punch). Could Intel simply be acting as a PowerPC foundry? Or could it be designing its own PowerPC spin? Hey, blue-sky thinking here, people……

Contined with 'Apple On Intel, The Final Chapter (At Least For Now)'….