It seems that a lot of people were confused about what I wrote about the AMD-Intel lawsuit yesterday, and even more confused about the subject as a whole.
The basketball analogy seemed to confuse most. The point I was trying to make is that there have been other reasons over the years besides Intel cash why big OEMs would shy away from AMD. Some years back, platform stability was a factor. More recently, delays in bringing out Hammer, and being able to make lots of them on demand has no doubt been a big factor in big dog OEM-land.
If you’re building millions of computers a year, it doesn’t matter if Product A is better than Product B if you can’t get a steady supply of Product A. That doesn’t matter to a person buying a single CPU; it matters a lot to someone responsible for having multi-millions of computers getting built.
One of the core points of the article was that the main source of Intel competitive power is that big OEMs can’t currently give all their business to someone else.
This is why Intel can play games like “if you buy more than X AMD processors, you don’t get any rebate on any of the Intel processors you bought,” which can make getting even free AMD processors for some machine more expensive overall than buying all Intel and shunning AMD.
At least some out there have said that I didn’t understand this when I said AMD would only need a 15% discount to match Intel incentives. Perhaps I should have added, “provided AMD could supply a particular OEM with most or all of its processors.”
It’s because AMD can’t promise to fully supply an OEM that Intel can play these kind of games. The threat of no Intel rebate only works when an OEM has to buy most of its processors from Intel. It vanishes if a big OEM could start buying most or all of its processors from AMD at say, the 15% discount I mentioned.
In other words, if Intel threatens an OEM with withholding a 10% rebate on 80% of a company’s CPU purchases, that’s a killer threat for somebody who wants to buy 20% AMD processors, but only so long as the customer is stuck buying 80% of its processors from Intel.
If the customer were able to buy 80 or 90% or 100% of its processors from AMD, though, Intel couldn’t make that threat stick any longer. The customer would just drop Intel and get its CPUs from AMD. That’s what I meant by AMD only needing a 15% discount; that’s all they would need if they could completely supply an OEM with chips. It’s AMD’s current inability to completely supply a big dog that puts them (and the OEMs) into this situation.
A lot of people seem to think that it’s illegal for a company to be a monopoly. Not so. There’s nothing illegal about simply being a monopoly under U.S. law.
What can be illegal is how you go about becoming a monopoly (i.e., mergers to become one) or what you do to compete against your competitors if you are one. Monopolies can’t engage in certain practices which are allowed for those companies that are not monopolies
To win an antitrust case against a company is a two-step process. First, you have to prove that the company is a monopoly. Only after that point is settled in your favor can you complain about the company’s practices.
I’m not even going to pretend to guess how a court is going to rule on whether Intel is a monopoly or not. To make a long story short, it will probably be a close call, one way or the other.
Even if Intel is found to be a monopoly, keep in mind is that different places have different antitrust laws. It’s quite possible that Intel could be found innocent in the U.S. and guilty in the E.U., simply because antitrust provisions are tougher in the second. That’s just what happened with Virgin Atlantic when they sued British Airways for certain practices. They won in the EU (see Virgin/British Airways, Commission Decision, Case IV/D-2/34.780, OJ L030,  4 CMLR 999, ¶¶ 108—11), but lost in the U.S..
Just understand that this isn’t a slam-dunk case, either way, and anyone who tells you otherwise, well, they need to do some more homework.
What AMD would like to do over the next three years or so is double their marketshare, from the current one-sixth to about one-third.
This means ripping away, permanently, about 17% marketshare from Intel, or about 20% of Intel’s business. AMD has never come close to doing anything like that even for a year, even when their product had the same or greater margin of superiority over Intel’s.
If you want to say that’s because of Intel’s nasty practices, well, AMD has certainly lost sales because of it, but if Intel were ordered to stop all this, would AMD sales would suddenly double as a result? There’s good reason to doubt that.
Whether sales double or not, AMD nonetheless will have to pay back most of the cost of the additional fab capacity they’ll have to build. This is a pretty indebted company as is; it will be a lot more indebted after such an expansion.
Yes, both the local and federal German government did help AMD out with Dresden Fab 30 and 36, but only a small portion of the fab cost came from them in grants. Much more came in the form of loan guarantees (i.e., if AMD can’t pay a loan, the government does).
Given economic conditions in Germany, and AMD’s own economic status, the governmental authorities there at the least may not be as willing to take on additional risk as they have in the past.
A very good economic reason for that would be that if Intel doesn’t change its investment plans and builds assuming the usual 80% marketshare, we’re going to have a big CPU glut in 2007-2008. We may have 15-20% excess capacity by then, and if Intel isn’t forced to rollover and play dead, both companies will likely find themselves in a price war just around the time a lot of loans start coming due.
This may end up being no big deal for AMD if Intel is found guilty as sin and has to fork over a few billion dollars, then be leashed and changed. Unfortunately for AMD, that’s just about the only scenario where it’s no big deal, and even if Intel ends up with that kind of judgment, they’ll appeal it to the very end, so AMD won’t see any money at the time when it might need it.
To pay off its loans, AMD is going to have to start making a lot more money from CPU than they are currently in a few years to be able to pay off the prior fab loans, and finance lots of future fabbing, maybe five-ten times more than they are now. If they can’t sell most of their capacity at fairly reasonable prices, they’re not going to be able to do that.
That’s what makes all of this so risky. It’s not the lawsuit itself that’s risky; it’s this assumption by AMD that it will just beg and borrow a few billion more, then almost painlessly double in relative size, taking it out of Intel’s hide, mostly due to siccing the government on Intel. It’s not the suing part that’s risky or questionable, it’s the painlessly doubling part that’s the problem, even if they win the case.
In other words, the combination of building much more capacity and suing is a high-risk breakout move, a move to get AMD up to being Pepsi to Intel’s Coke. There’s certainly the possibility of a huge upside if AMD can, one way or the other, sell 80-100 million CPUs a year at good prices in a few years. It’s conceivable that AMD could end up with that even if it loses the lawsuit.
On the other hand, if AMD doesn’t end up selling 80-100 million processors at good prices in a few years, it could face bankruptcy even if it wins the lawsuit.
Whatever happens, it’s not going to be simple or easy.