Apple rumours and realities

It's not possible to make money selling iBooks in which thewholesale CPU costs amounts to more than one third of thetypical selling price
Written by Paul Murphy, Contributor
Quite a number of pundits have gone out and predicted the appearance of Intel based Powerbooks from Apple this January - but it isn't going to happen. In fact what I expect is either, or both, an Intel based Mini and/or some Intel based iBooks. Beyond that, however, Intel is just not going to happen for Apple next year - no Powerbooks, no iMacs, no Workstations, and no X-Serves.

To see why consider Apple's strategic trilemma:


  • there are both market (demand related) and marketing (selling related) reasons for trying to get Intel based iBooks and Minis out early;


  • but in the short term costs are higher then expected, CPU availability is limited, and the opportunity to realize performance gains on professional level gear depends more on bad software than good hardware; and,


  • longer term prospects are clouded because Intel is failing to deliver on its pricing, volume, and performance promises.

The case for getting iBooks out early is very strong and comes in three parts:

  • first, schools have long been a sustaining market for Apple but districts like to go shopping during the spring and most won't consider something that's not at least nominally shipping in that period.


  • second, there are lots of Apple loyalists in districts where an early "Intel Inside" label could let them sway big decisions Apple's way -to produce not just sales but that all important validation for the Intel decision; and,


  • thirdly, the iBook has traditionally been on the low end of the performance scale, so a switch to even the older 32bit Pentium M should produce excited reviews claiming speed-ups ranging from 30 to 100 percent - again ratifying the Intel decision in the battle for public opinion.

Unfortunately there are two big problem sets: one is pricing, the other availability.

Of these, availability issues are the least flexible but also the easiest to finesse, at least in the very short term. Apple needs the new instruction set extensions promised for the "Yonah" architecture both for performance and to support its "best efforts" hardware copy protection on MacOS X. Unfortunately, "Yonah", even in its first 32bit incarnation, isn't ready and its full implementation successors, "Woodrow" and "Merom", keep getting further and further behind schedule.

There are at least two "big lie" workarounds, but both combine high risk with few off-setting opportunities beyond the simple reality of getting at least some product out the door. The first of these is just to go ahead and announce availability knowing that in reality only demonstration quantities can be shipped. The second is to go with the older Pentium M in the early production machines knowing that the later "Yonah" machines can be spun as a second generation and the change over presented as further validation for the Intel strategy.

Pricing, in contrast, is a simple rocks and hard places choice.

Right now a $999 list price iBook sells in volume for something closer to $699 and earns a small but positive margin for Apple. The PowerPC G4 in that machine has a typical volume price of around $72, or about 10% of the selling price for the machine. In comparison people like Asus, Quanta and Hon Hai Precision (who make Dell, HP, and IBM gear) pay Intel on the order of $240 per unit for the two year old, 32bit, 1.8Ghz Pentium M predecessor to the "Yonah" line.

It's not possible to make money selling iBooks in which the wholesale CPU costs amounts to more than one third of the typical selling price.

Something has to give here: either Apple has to significantly raise prices in public or cut everything else to the bone. Unfortunately raising the price significantly isn't acceptable in the market, so if Apple wants to do sell Intel what they'll have to do is accept lower product quality in other components and a higher DOA rate out of the plant gate, settle for a two year old, 32bit, chip without the new media instructions characterising the "Yonah" architecture, cut its plant gate margins on the product, and reduce discount levels available to volume buyers.

In other words to hold the list price constant on the iBook in the face of such a massive cost increase for the CPU, they'll have to reduce both customer discounts and their own margins, take a big downstream hit on component quality, and give up on CPU level MacOS X authentication.

Other companies have done this: in fact this is exactly the recipe that turned Dell into a company that just puts the bruises on the bananas, gutted Compaq, and led to years of losses at IBM's PC division.

I should note, however, that one of the interesting rumours about the hurry up and sign something deal Apple worked out with Intel when they finally gave up on IBM has it that Apple expected to get subsidy pricing from Intel on some processor lines - unfortunately the ones that aren't ready yet and may simply never make it to volume production.

The situation with the Mini Mac isn't that different. Nobody can make money on a $499 computer in which the CPU eats up nearly 50% of the total retail price. - Microsoft, which is thought to pay $106 for the six thread, three core, 3.2Ghz PowerG5 derived processor in its xBox, is also thought to subsidise unit sales by about $200. Apple can't do that, especially with CPU prices that start over $200, but faces the same strategic pressure to get its product into people's living rooms.

So what can they do?

The obvious answer is to stick to PowerPC for another generation - pushing the first Intel products into 2007. IBM has a low power (13 Watt) G5 that would be a big winner in new PowerBooks, and Freescale's 8641, a dual-core PowerPC G4 with integrated system logic and four Gigabit Ethernet media-access controllers, offers exactly the price/performance combination Apple needs to give both the iBook and Mini big performance boosts without changing retail price or cutting their own margins.

Doing it would reward Apple's most loyal customers and those developers who put real effort into working with the PowerPC while leaving the company with a bigger market share and time in which to make a considered long term CPU decision.

Sadly, however, good sense isn't that likely to break out where personalities are on the line -instead we're more likely to see Apple spend money along with both customer and developer loyalty on building enthusiasm for Intel solutions that are virtually guaranteed to eventually gut the company financially.

Editorial standards