I should make it clear that I don't think that Intel is cheating or practicing explicit fraud in their accounting. But one set of GAAP rules is used to cover every business in the country, and those rules don't work very well for non-standard businesses. For instance, Indianapolis coke had a big capex cost in 1908-1909:
Twenty-two acres of land at the Utility's present Prospect Street plant site were purchased and a water gas plant was hurriedly constructed in 1908. Water gas production then commenced at the end of March, 1909. Two coke oven batteries of 25 ovens each were also constructed. In November of that same year coke oven gas production began and has continued at that site to the present
indycoke.com
Coke gas isn't very important these days, but the metallurgical quality coke produced in these ovens continues to be in high demand from local steel mills.
Much of the equipment bought then is still in use (7x24x365, as a matter of fact. It takes a number of days to bring the ovens up to operating temperature, so they are never shut down). GAAP allows only limited differences in the handling of that expense, and Intel's expenses to buy stepper-scanners and polishers that lose most of their economic value in a few years. Admittedly, these are the two extremes of capital investment to be dealt with in GAAP, but that's just my point - Intel, and AMD, (and NT!) are extremes in terms of accounting, and need to handled very carefully. Blindly following GAAP, isn't enough, and NT and Lucent are great examples why.
AMD finds ways to expense capex about as fast as it occurs, while Intel blithely accepts that its capex costs are often significantly under-expensed in GAAP, assuming a "not my problem - let the investors try to figure it out" stance.
With Intel touting its "profits during the slump" nonsense, it makes AMD's performance appear, by contrast, less spectacular than it has been. |
