I was somewhat surprised at Intel's somewhat stronger than expected earnings this week. While I didn't expect a bad quarter, and was in fact leaning slightly to the long side on the stock, I didn't expect that much.

Worth noting that Intel is conservative in its guidance, and suggestive that things won't last indefinitely.

Even more suggestive of this is the fact that inventories and recievables are up as well, despite relatively flat revenues. And of course, they note that part of the reason for their great margins is that they were able to sell "previously written-down inventories."

This inventory game is one that Intel has always played, even a decade ago when I worked there. It's become more pronounced in recent years though, a process I believe is attributable to Craig Barrett's greater tolerance of such things. (Andy Grove, for better or worse, was intolerant of just about everything...)

The game is simple. Crank out as much product as you can, which in a fixed-cost business like semiconductors means you end up with a lower average cost per product.

Sell only that product that can be sold at the desired price, which keeps margins high. Report the difference between the (low) cost of production and the (high) cost of sales as the "margin." Ignore the fact that you didn't really sell all the inventory at that price, and some of it is still sitting around.

A quarter or two later, write down the value of the unsold inventory to some number well below the cost of production in a "one time adjustment." Sell the inventory cheap. Your margin represents the amount of cash you're able to bring in less the cost of goods still on the books. Since that cost was written down to an arbitrarily low number, margin still looks good.

The write-down amount, which would represent a real impact to margins if included, is excused by the company and its fans on Wall Street as being "irrelevant." After all, the company did say it was a "one time" problem, right? Even though they do this repeatedly. Few bother to go back, add back the cost of the writedown and try to figure out the real margins.

In short, it is clear to me that the "margin" reported by Intel really isn't. It's an accounting obfuscation that fails to actually capture the difference between the real cost of goods and the real value of the products created in the marketplace.

There is another thing going on at Intel though, and it has clearly thrown off some of our thinking lately. The growth is all in Asia. To some degree this is US end-demand that is being satisfied by Asian product manufacturers, but to some degree it is local demand. Virtually all my contacts in the end-demand world are in the US and Europe. Clearly, I need to adjust my thinking and analysis to include more feedback from and observations of Asian players.

Bill Fleckenstein surmises that some of the Asian demand is merely channel stuffing. He may be right, but at this point I don't have any real way to know. Anybody with thoughts on the matter would be welcome.

Intel has been a nice bet this week, up almost 5% since our entry three days ago. But I have little long-term conviction about the name and will likely sell before the weekend.