25 May 2006

Apple's Intel Choice Looks Inspired

Apple's announcement that it was moving it's Mac line to Intel chips last year was met with shock and some disbelief. Those that thought the move was long overdue criticised their selection of Intel - they should have gone for AMD who have been beating the pants off Intel in the last few quarters. There were also a few that insisted the PowerPC was still the right choice. A few of them have even being trying to resurrect the hope of the PowerPC being a future contender by clinging onto stories such as this "exclusive" from the Register last week which got some attention. Essentially this introduced the fact that Apple had another choice other than IBM or Freescale - a small start-up called PA-Semi. While I'm sure Apple looked carefully at this, it is hard to see how the company offered a viable option for Apple in a suitable timeframe. While one version of a chip would have been shipping in quantity early next year (theoretically), that would not have met the needs of Apple across the range of machines. And quite where Apple would be now with a G4 line of laptops is hard to imagine.

But back to the Intel/AMD argument. Dell's recent selection of AMD is being used by some to reinforce the argument about Apple making the wrong choice. But it is increasingly clear that Apple's timings on selecting Intel have been either inspired or fortuitous. The Core Duo chips used in the all the consumer models have been very well received indeed by the mainstream PC press. AMD has only recently been able to respond with a Dual core Turion model suitable for laptops. And, within the next two-three months, Intel will have introduced the next generation of Core chips - the Core 2 architecture. This architecture is the basis for a whole range of chips from the "Merom" versions for laptops, through "Conroe" for the higher end desktop to "Woodcrest" for the server market. All of these chips are fabricated using 65nm technology (AMD is still on 90nm until the end of 2006). This article from Ars Technica is a great summary of the situation with Woodcrest and links to three detailed reviews and benchmarks next to older Intel chips and current AMD leaders. Merom has already been reviewed well in other places. These chips are going to catapult Intel into a lead that AMD will struggle to match for the next 18 months or longer across the board, with the exception of high end 4-way or more servers. The chips perform well both on absolute performance but also when considering power usage, we are witnessing one of the biggest improvements in chip performance for many years. We will no doubt see Woodcrest in replacements for the Powermac line, as well as LV versions to be used in the XServe replacement. It will be interesting to see where the Conroe versions end up (if at all). Will we see a new model using these or will they sit in the iMac?

Of course, there were other factors in Apple's choice. While Apple represents a tiny fraction of Intel production, AMD might have actually struggled to meet Apple's demand which would have necessitated an increase in production of perhaps 25% or more. Intel seems even to have actually exceeded its own expectations of production - Woodcrest is coming in ahead of time, and it seems to have had no problem shipping quantities of Core Duo chips so far.

While there is little doubt that AMD will bounce back, and that financially at least, they are still causing Intel all sorts of problems, all of this represents the sort of healthy competition that Apple can now benefit from - in terms of price, performance and choice.

Of course, we may never know whether Steve went Intel because he wanted to offer machines that could also run Windows easily, or whether he was genuinely seduced by what was being offered and saw the Windows opportunity as an extra benefit. But with the choice and timing (any earlier and Intel looked quite sick; anything later would have missed building on the iPod halo effect) it is increasingly looking like he's made one of his best ever business decisions.

No comments: