Apple's Intel decision – does it necessarily mean x86?

You might be surprised to find a story about computers on a blog that traditionally focuses on atomic energy. The “tech” field, however, provides some interesting models that can be useful to people interested in the continuing development of practical atomic technology.

One of my favorite companies is Apple Computer. (Disclosure: I own a number of Apple computers, several iPods, iSights, an AirPort and some Apple stock.) I love the story about a couple of young guys who started their company with few resources other than a deep love of tinkering and a desire to produce useful gadgets for their friends. I am impressed by the way that the company has matured and learned from its mistakes. Finally, I am enamored by their “systems” approach – Apple does not produce isolated components, they produce complete systems that include hardware, software, case design and, perhaps most importantly, the ultimate user. As a systems technologist (I earned my MS at the Naval Postgraduate School in Systems Technology) their products provide me with daily inspiration.

Intel is another company that has provided inspiration and business models worthy of emulation. That company has aggressively improved its products for more than three decades, never content to rest on its previous successes. I am not such a fan of their business practices, but they make great products and are moving towards making great systems themselves.

Apple’s recent announcement that they were changing their microprocessor supplier from IBM to Intel has provided the grist for a lot of conversation in many forums, but it made me think about something that many do not seem to remember. Perhaps there is an advantage to getting older.

In the mid to late 1980s, Intel engineers recognized that microprocessors using RISC (reduced instruction set computers) architecture had the potential to eventually outperform their CISC (complex instruction set computers) based x86 line of chips. At the time, the “X” equalled a 3; the 486 was still under development. Intel put a lot of resources into developing the i860, which was their first RISC chip. There was a vocal faction in the company that expected these new chips to become the basis for the future.

Another large faction within the company recognized that there was still plenty of room for improvement in the x86 line, even if RISC based chip development might ultimately follow an “S” curve with a higher end state. (See, for example risc vs. cisc. For an understanding of S-curves in technology development see The Innovator’s Dilemma.)

Since moving from CISC to RISC would require its customers to develop entirely new software, the change would disturb the installed base, one of Intel’s primary business advantages. In April 1989, after Intel successfully introduced the 486, Andrew Grove focused his company on producing ever improving x86 based microprocessors. In 1996, he wrote “I shake my head about how I could have even considered walking away from our traditional technology that then had, and still has phenomenal headroom and momentum.” (Heller p. 63)

Interestingly enough, Intel never stopped its work on RISC processors. Its advanced line of XScale® processors – “designed to optimize low power consumption and high performance processing for a wide range of wireless and networking applications and rich services” – (ref: Intel XScale® Technology) use the RISC architecture and provide entry into markets that cannot be served by the power hungry and relatively expensive x86 processors. Of course, the software written for these chips cannot run on x86 based PC’s, but Intel sells them into different markets so that they are not cannibalizing the installed base.

I have carefully read the text of the announcements from both Apple and Intel; I cannot find either of them specifically stating WHICH Intel chips Apple will be buying. It would seem consistent and logical for Apple to choose the chips that are based on the architecture most like its current architecture and for Intel to sell them chips that best meet their design needs. A key point to remember from Job’s speech is that the performance per watt was a key factor in the decision to switch, and it appears that XScale® chips are designed for low power consumption.

From Intel’s point of view, selling RISC chips to Apple provides them a new market with plenty of room for growth. From a technical point of view, I would guess that the x86 “headroom” that Grove mentioned in 1996 will be gone by the time that Intel based Macs hit the market. After all, Intel doubles its chip’s capacity every 18 months so ten years is almost 7 generations. If the company is going to continue its quest for ever increasing performance – using several different measures of effectiveness – it would seem that the RISC line is less risky.

Here is one more quote from Heller’s interesting book that supports the logic of this theory. The context of the quote is that Heller is describing Grove’s theories about recognizing a strategic inflection point, when your business is about to change or be changed. Grove suggests that business leaders ask themselves three questions, any one of which might help them identify when change will happen. The pertinent question in this case is:

Is your key “complementor” about to change? If the company that mattered most to the business seems less important, that again may signal fundamental change. (Heller, p. 80)

The momentum and leadership in personal computing is shifting; Apple and its products are far more important than they were ten years ago. Pundits are paying a lot of attention to iPods, probably because they are so visible, but there is a lot happening with OS X, Xserve, iLife, iWork, iTunes cellular phones, and probably a whole host of products that we do not yet know about. Microsoft is still very important to Intel, but more as a cash cow than as the basis for the future.

Within the next few days, I will try to get back to the reason that I introduced this topic into Atomic Insights. It has a lot to do with a decision facing key nuclear plant vendors that involves changing technology models, cash cows and room for growth.

Ref: Heller, Robert; Andrew Grove: The Innovator Whose Methods Supercharged the Silicon Revolution; Dorling Kindersley, London 2001.