We know the overall effect on the CPU price war: Intel makes less money; AMD loses a bunch. But what impact has it had on the average price per CPU sold in the various niches?
This link provides that kind of information:
Over the last ten years, when Intel has done well, it has had an ASP of around $150. AMD has usually needed an ASP of about $90 to make a decent profit; it has been as low as $60.
Maybe the best way to illustrate the difference between the finances of AMD and Intel is to tell a little story:
Once upon a time, there were two stores named AMD and Intel selling widgets. Under normal circumstances, let’s say that if you add in all the direct and indirect costs of the widget, it costs AMD $10 to make the widget, and they can charge $11 for it.
Over at the Intel store, the widget costs $15, but Intel can charge $20 for it. For whatever reason, good, bad or indifferent, people are willing to pay more for the Intel widget, so while Intel’s widget costs them a good deal more than AMD’s widget, Intel makes a much bigger profit than AMD because it can charge a lot more for its widget than AMD.
One day, there is a price war. which starts due to a combination of Intel wanting to dump an old widget line, and AMD feeling that they need a bigger share of the widget market to stay in business long-term.
To make a very long story short, Intel figures out how to reduce its cost per widget from $15 to $13, and cuts its price from $20 to $15-$16. They’re still making money, but not as much as they used to.
At the AMD store, they find that it costs a lot more money to make a lot more widgets and make them more appealing. Not only did they have to empty their piggy bank, but their average cost has gone from $10 to $11. To their horror, while they find themselves paying more to make widgets, they now find that they can’t sell their widgets unless they sell them for $7.50.
This is roughly where we stand now.
The people at the AMD store know they can’t keep selling $11 widgets for $7.50 and stay in business, no matter how many more they make and sell. So they’re getting a new widget line ready, which they hope to sell for $15, $20 or maybe even $25.
They’ve been very hush-hush about it, but it looks like their having problems a) making widgets for which people will pay much more money and/or b) quickly shifting their production to start making and selling a large proportion of these higher-priced widgets rather than the $7.50 specials.
The story is much different at the Intel store. While the managers and especially the investors in the Intel store aren’t too happy about making just $2 a widget rather than $5, they can keep doing this far, far longer than AMD can sell $11 widgets for $7.50.
In the very long run, the people at the Intel Store probably do need to make more money, if only to pay for lots of those extremely expensive widget factories, so they’re looking at new ways to do that, one of them being a $5 mini-widget that will cost just $2 or $3 to make, and selling lots and lots of them.
Maybe that works, maybe it doesn’t, most likely it works somewhat to pretty good, but the Intel Store will be making big widgets for a long time to come. How can they make more money from them?
Well, no matter what they do, it will be a whole lot easier for the Intel store to make more money from anything they decide to make if the AMD store runs out of money and stops making widgets. Then they can go back to charging $20, maybe even $25 for the big widgets.
Maybe it’s worth making just $2 a widget for a while longer when there is reason to believe it might make the AMD Store go away.
It’s not quite as easy as that, and making the AMD Store go away would cause new problems for the Intel Store, but it sure is tempting.
What’s the end of the story? Nobody knows yet. Fairy godmothers might show up; so might more big bad wolves, maybe both, maybe even a fairy godmother wolf.
I doubt everyone will live happily ever after, and I’m not too sure everyone will live much longer, period.
Tomorrow, we’ll look at these numbers from a different angle.
Be the first to comment