Debunking Dvorak With Napkin Math For OS X On Intel

by , 1:10 PM EDT, April 8th, 2003

Let me get this out of the way first. I think John Dvorak is great. I love his Macintosh articles the most because they tend to piss everyone off. I also like them because they make me think a lot more than articles written by apologists. However, as often as I enjoy his articles, I also find myself disagreeing with some of their content -- content I suspect is there to incite thought and reaction more than confidence in their propriety. So in that spirit, I greatly enjoyed his latest editorial. In that piece, he lays out what he thinks will be the time table for Apple moving Mac OS X to Intel, as well as how he thinks it will affect the market.

As is the norm with his writing, it got me to thinking. In this case, how realistic is a move to OS X on Intel hardware by way of licensing to clone makers? So I fired up a spreadsheet and did some napkin math to see what it would take. Here are the results:

I obtained Apple's SEC 10-K filing and found its Net Sales, Cost of Sales, Gross Margin, and overall number of Macintoshes (Mac Units) sold for 2002, as reflected in the table above. Using some rough math, Apple makes about US$500 per Macintosh sold. Now of course they make more per unit on high-end machines and less on the lower end machines and the Gross Margin figure contains revenue made from software and service sales. Be that as it may, the US$500 figure will suffice for some rough ball-parking. Incidentally, Apple's 2002 10-K report shows that their Gross Margin percentage was 28%, 23% in 2001, and 27% in 2000; pretty darn good. Finally, the last I read, Apple enjoyed a 2.4% global market share.

So after getting the above figures, I was left with the question of how many copies of OS X for Intel would Apple have to sell to maintain its gross margin income of US$1.6 Billion (i.e., so that Apple more or less maintains its current size)? The short of it is this: If Apple would make about US$50 per copy on sales of OS X for Intel (i.e., after costs, discounting to OEMs and letting retailers have their share of the US$129.00 retail price), it would have to sell about 32 Million copies (you can see estimates based on other per Unit Margin on Software values in the above table). Apple would have to sell roughly 10 times as many copies of software as they do Macs to maintain gross margins. If you apply this linearly to market share, Apple would have to obtain, roughly, a 25% global market share to maintain gross revenues.

Now before I go any further, a few caveats as I have taken ridiculous liberties in extrapolating the above numbers. First, of course you don't simply extrapolate the market share figures linearly as there is already a large base of Intel machines in place that are potential buyers of OS X for Intel. Obviously, that would lower Apple's overall market share requirements in maintaining gross margins. Second, if Apple became a software only company, it wouldn't require the same levels of gross margins as it wouldn't require as much infrastructure (although if they continued to both develop hardware and software, the increased requirements in support might require it to actually increase gross margins). Third, Apple wouldn't necessarily become a software only company over night or ever; it would supplement gross margins on OS X for Intel with hardware sales, which would lower the market share requirements. Fourth, IDC's numbers do not account for many types of sales and are not complete indicators of market share. Fifth, market share numbers for hardware do not necessarily map well to the realm of software; and finally, many other caveats apply, but again, this is napkin math.

Significant caveats aside, the numbers provide some cold water in the face of flights of fancy. Should Apple simply go to the stage of licensing its software with no controls, it needs to balance every Macintosh that is not sold due to cannibalization caused by OS X for Intel with roughly ten copies of OS X for Intel to maintain gross margins. Of course to some degree, were Apple to concentrate more on software (e.g., by developing an office suite, iOffice, and charging for more of its iApps), it could increasingly make more money on applications to help increase on the gross margins front. What is interesting to note, if Apple were to sell some application software for every few copies of OS X, it would require more modest market share thresholds, i.e., if it were to make a US$100 per copy of OS X for Intel (e.g., US$50 for OS X and the sale of some iApps or a sale of iOffice for, say, every 5 or 6 copies of OS X), it would require only 12% market share to maintain gross margins. If that is combined with some sales of its premium hardware, a market share of 10% might suffice. Regardless, it is highly speculative and reckless to assume that Apple can instantly obtain this 10 to 1 ratio to forestall lost gross margins on Macs with software sales to Intel/clone PC owners.

More realistically, if Apple even bothers to go to Intel processors, its best bet is to test the waters with its own Intel PC first. I believe John Dvorak is very much right that Intel has some reservations with Microsoft and is not thrilled about relying on it alone to drive sales. But using Intel processors does not necessitate licensing Apple software to clones. In putting together its own Intel machine, Apple could make it every bit a Mac in design, except it would have a faster Intel processor and memory. By making its own Intel Mac, Apple can put in a decryption key into a ROM chip and program OS X for Intel not to boot unless it obtains the proper decryption key. This would allow them to maintain fat margins on machines with faster processors while controlling and preventing any significant cannibalization to the clones. Moreover, being the "Lexus" of computers, Apple might enjoy increased sales of Intel Macs to Windows users for people that need to run Windows but want the high quality experience of Apple hardware designs; an Intel Mac would easily be able to dual boot into OS X or Windows alone.

By the way, an Intel Mac might (allow for and) require some interesting emulation options (e.g., "red box," rootless window, Windows emulation could be implemented somewhat like the Classic layer with the help of the following types of technologies: Lindows, native versions of Virtual PC, or Wine). Even if Apple were not able to provide Motorola processor emulation to accommodate for Classic applications (and there is no reason to believe Apple could not), most OS X applications have been Carbonized and OS X provides an infrastructure that allows for fat binaries that would run on either Intel or PowerPC chips natively; such fat binaries would allow for a smoother transition to Intel if needed.

Assuming the Intel Mac trial balloon went off successfully, Apple could then license the requisite decryption chips to other manufacturers (e.g., Dell, Gateway, etc.) for limited trials and test how much such sales would affect its bottom line. This would allow for a much greater control to effects on its gross margin than letting loose OS X into the lands of Intel clone wars. Perhaps as it ramped up its market share, Apple might eventually be able to support unfettered licensing of an Intel version of OS X. However, as it stands, John Dvorak's and others' hopes for OS X on Intel seem to be cries for Apple to tilt at windmills.

John Kheit is an attorney. Please don't hold that against him. This work does not necessarily reflect the views and/or opinions of The Mac Observer or even John for that matter. No assertions of fact are being made, but rather the reader is simply asked to consider the possibilities.