Some thoughts on the economics of personal computers
Published on June 19, 2005
By
Leauki
In
Had Intel suceeded in establishing Itanium on the desktop, as they originally planned, Itanium chips would probably be cheaper than they are now.
Of course, once the chip is established as the new standard, like x86 before, the price would have gone up, since users need the chip as much as the operating system. The result would have been a monopoly of standard microchips compatible with the Intel Itanium just as there is a monopoly of standard operating systems that are fully compatible with Microsoft Windows. Patents and copyright guarantee the monopoly, just as it was intended.
Microsoft, of course, knew about that danger. It is in Microsoft's (and every operating system vendor's) best interest for processors to be commodities (aka cheap and interchangeable). Thus Microsoft made a few moves to make sure Intel and HP could not succeed (while still supporting the Itanium chip as promised).
Microsoft knew about this danger in the late 90s and acted accordingly:
1. Keep everybody 32 bit until the arrival of a commodity 64 bit architecture.
2. Support Apple with Mac OS applications to keep other processor architectures around.
3. Move Xbox to PowerPC.
4. Support the commodity 64 bit architecture when it appears.
Intel also knew and were thus working to make operating systems the commodity.
1. Start and support a Linux-port to Itanium.
2. Try and convince software vendors to port operating systems to Itanium (including AIX).
3. Get HP to port HP-UX and VMS to Itanium.
4. Get Microsoft to offer Windows for Itanium.
The perfect situation for Microsoft is many competing vendors for all products that complement Windows: application software, games, utlities, hardware including graphic cards, microprocessors etc..
Thus Intel and AMD (and Cyrix and Transmeta etc.) producing what is understood as a commodity chip is what Microsoft want. But Intel alone offering the new standard chip without competition, that is as bad for Microsoft as Microsoft's monopoly (again, in the market for fully Windows-compatible operating systems) is for everybody else.
Note that this "everybody else" is makers of computer components (including software), not customers. Customers already pay as much as they will for a computer and this pricepoint is not affected by exactly which component of the computer is a commodity and which is not. An Itanium-based computer with Windows would, in the long run, cost as much as a Windows computer using a commodity CPU, namely as much as customers will pay for the entire solution. The only difference is the percentages of the money Microsoft and Intel or other chip makers receive.
If the processor is a commodity, Microsoft's percentage will be higher. If the processor is not a commodity (say an Itanium standard which only Intel can supply products for), Microsoft's percentage will be lower by exactly the amount Intel have gained.
Customers directly profit from commodity components when every single required component is a commodity. If the operating system as well as all the hardware is a commodity, prices for the entire system will fall, because none of the suppliers has a monopoly on anything and thus none of the suppliers would be in the position to earn the difference between cost of production (the price of a commodity) and what customers are willing to pay (which can be a lot more). The amount paid in excess of what commodity components would cost is what economists refer to as "rent". In this case the rent is a "monopoly rent" since it exists due to a monopoly power. This is similar to the reason for why an apartment in a nice neighborhood costs more than the same apartment in a bad neighborhood, in case you have ever wondered.
There are degrees between the two extremes. For example the operating system could be replaced by something that is not exactly the same thing but sufficiently like it. This would bring the price of the operating system down, i.e. commoditize it a bit. Linux has that effect. OS/2 did. Novell DOS did.
Sometimes no operating system at all can replace Windows. The customers in question don't need it and replace it with anything. In that case the operating system becomes a commodity, and I think one might notice that in these cases a computer with Windows and without Windows will cost the same, because the pricepoint is not affected by the exchange.
Think "Corn Flakes with Milk".
If both Corn Flakes and Milk are commodities, "Corn Flakes with Milk" will cost as much as it costs to produce them, say the amount C+M, with "C" being the cost to produce Corn Flakes and "M" being the cost to produce Milk.
If both Corn Flakes and Milk are produced by only one source, "Corn Flakes with Milk" will cost X, with "X" being the maximum amount customers are willing to pay for "Corn Flakes with Milk". This is the "pricepoint" referred to above and in extreme cases (if there is absolutely no other food at all) it will be infinite. [0]
If Corn Flakes are a commodity and Milk is produced by only one source, the product "Corn Flakes with Milk" will also cost X, with one of the Corn Flakes makers making C and the Milk maker making X-C (which is more than M).
If Corn Flakes are produced by only one source and Milk is a commodity, the product "Corn Flakes with Milk" will again cost X, with one of the Milk makers making M and the Corn Flakes maker making X-M (which is more than C).
If Corn Flakes are produced by only one source and Milk is a commodity and Corn Flakes can be replaced by something that is not quite the same as Corn Flakes but a somewhat acceptable replacement, say Rice Flakes, which are produced by only one source, the product "Rice Flakes with Milk" will make Y, for which we know is true: YM and the Rice Flakes maker will make Y-M and M will be, as above, the cost of producing the Milk.
And if a customer only wants Milk and doesn't care about Corn Flakes, even if only one source produces Corn Flakes, the customer will pay for "Corn Flakes with Milk" the same amount he would pay for just Milk, whatever that costs. Thus, for this customer, Corn Flakes would be a part of the commodity "some product I don't need", which is, of course, the ultimate commodity (the price is below even production costs at zero).
Thus Intel's Itanium chip would have cost less than what a PC costs but the price of it would have been a greater percentage of the price of a PC than Intel's chip is today.
And Microsoft certainly did not want that.
Thus Itanium now competes against SPARC and POWER in the server market where CPUs are expensive and operating systems tend to be commodities.
And that is, simply put, why Mac OS was ported to Intel x86 but not to Intel's IA-64 architecture.
Microsoft did not want it to happen. And Apple must see it the same way.
[0] This is not quite true, or at least it is possibly not quite true. The monopolist might not be able to sell anything if the price is too high because all customers wait for the price to fall and nobody wants to be the first buyer. The price will eventually fall because the monopolist can make a greater profit by selling to more customers. In order to sell to more customers the monopolist has to offer the product for a price more customers can afford. Customers can know that and thus don't want to be the first buyer. And the cycle repeats.