Apple used to make pricey computers using chips not found in everyday PCs, and needed to charge (relatively) a small fortune because the components cost Apple more than vanilla boxmakers' parts cost them.
When Apple announced its future computers would be Intel machines and provided a roadmap for migrating applications to cross-platform builds using the Cocoa application frameworks, it also promised to save Apple from being in effect the party bankrolling an arms race against Intel for performance. Margins seemed to benefit from lack of need to re-invent the wheel. Apple could focus innovation where Apple had real fire, consumer software (it bought all the pro apps it sells, and its home-grown pro apps -- whether stand-alone software like Aperture or Software-as-a-Service offerings like its old email system or its new cloud computing initiative) haven't exactly demonstrated a flair for solid or reliable performance worthy of professionals' trust) and cosmetic design.
(In consumer software, though, Apple has some real success: folks seem happy to buy iLife upgrades, and iTunes has supported zillions of iPod users serviceably enough while helping Apple sell music at a faster clip than can #2 U.S. vendor Wal-Mart. Operating system upgrade revenue is significant in launch quarters to move the margins needle noticeably.)
By owning the operating system when it moved to commodity hardware, Apple gained a terrific position from which to compete: Apple could distinguish machines with Apple software and operating systems while offering compatibility with competitors' products, all while building machines for similar cost to those of competitors but being free never to pay a royalty to Redmond for every box sold. Even undercutting competitors on price, Apple could enjoy higher margins, because Apple would avoid the per-unit cost of an operating system license.
As Apple spreads its fixed operating systems costs across millions of units -- including cell phones and some iPod models -- the per-unit cost to Apple can become trivially small. Yet, as a major hardware vendor in the U.S., Apple can command the best prices on components used in every manufacturer's machines. The more machines Apple sells, the better its margins.
And that's before factoring in software and service sales. That's just the hardware business.
Apple apparently has an itch, though. Apple reportedly will start shipping custom portable computer hardware, involving custom chipsets or motherboard architecture, in order to achieve either size or performance objectives. Leveraging technology from the HyperTransport consortium, Apple may try to reduce latency or increase throughput between components and the CPU, though Intel plans addressing these issues with hardware to be released later in the year.
Why is the performance issue so critical? Does it significantly impact battery life, or other features of significant impact to mobile users? Why should Apple compete on computer hardware technology when it can simply coast along on the tech accepted by every other vendor?
Is Apple angling to rouse Intel to provide special terms to Apple to keep it loyal, by showing that its software architecture enables it to make snap decisions on hardware changes any time Intel is found wanting?
The bottom line is a lower bottom line: if Apple invests in competition to build hardware available as a commodity to competitors, Apple will be building niche gear at a relative premium at the expense of the margin cushion earned by its highly-leveraged operating system investment, but it will do so in a way that involves not only up-front R&D costs but ongoing per-unit production costs over the entire lifespan of the technology. If that lifespan isn't long -- and what is, in the tech world -- Apple faces retreat to commodity hardware or another round of R&D expense. Yes, Apple can certainly choose after it's designed its alternative between its own tech and the commodity parts, but it would still bear the economic burden of engineering the alternative.
Why bother doing this, when Apple's unit volumes surely enable it to play existing manufacturers against one another in an arms race to make a winning Apple supply bid -- and at no cost to Apple beyond seductive phone calls?
There may be a place Apple might productively engage in such research -- like the handheld market, where there is no standard platform from which to ride the train of commodity OEM part supply. I'm not sure the notebook market is the place for this, though I'm open to seeing arguments on the suitability of current tech to the market. After all, Apple did get Intel to engineer special packaging for MacBook Air parts.
So, what's Apple thinking? Is the custom tech for handhelds or tablets or some market that's poorly addressed by existing hardware, or is Apple just itching to wage war on multiple fronts again?