I’m getting some strange responses to this, including from some people who should know better.
It seems like a lot of them are just repeating the AMD party line on this, whether or not it has anything to do with what I said. Not saying at all that these are AMD-inspired; for some reason, plenty of people on the Internet defend their “cause” without any outside help at all.
Here’s some of the lines I’ve been getting:
“Grow up. AMD doesn’t have to come out with a 1.5Ghz processor just to make you happy.”
Maybe I don’t write well. I thought saying, “I said AMD should introduce Palomino, not introduce Palomino at a high speed” meant “I said AMD should introduce Palomino, not introduce Palomino at a high speed,” not the opposite.
Now where did I get this strange notion that Palomino should start at a lower speed? Why, I’m just repeating what AMD said. Even AMD agreed with this just a month ago. They must have thought it was a pretty good idea, too, since publicly stated that they were going to go into full production of Palominos starting at 1.2Ghz within a couple weeks. Seems like they were pretty far along in their planning to say something like that, don’t you think? Then something happened.
“Palomino is bigger than TBird, AMD will get fewer chips out of it.”
Yes, Palomino has a slightly bigger die size than the TBird. But didn’t AMD know that four weeks ago? After all, they just design and make the chips. The bigger die size didn’t seem to bother them at all a month ago, why the change in attitude?
“You don’t know anything about business, you stretch out product life when you can to save money and maximize profits.”
There’s a couple problems with this.
The first one is that this argument confuses product introduction with product speed. It’s a good argument for not having a 1.5Ghz Palomino now; it’s not a good argument for not having a lower-speed Palomino now.
If Palomino were just a souped-up TBird, with the same heat and power requirements, that would be one thing. But it isn’t. The lower power requirements and heat output all by themselves would justify their introduction, though a couple people almost said heat was a wonderful thing in defending AMD’s actions.
The second problem is that AMD is not alone in the CPU universe. There’s this company called Intel. They’re going to have
a .13 micron Willy at best by fall 2001, at worst by early 2002. Intel apparently is going to try to get a 2Ghz .18 Willy out sometime this year, the introductory .13 micron Willy will have to at least reach that speed, and quite possibly a good deal more. If Intel gets daring, we might see 2.5Ghz fairly quickly.
So AMD can’t dawdle on its own .13 micron processors. They plan on getting them out 1Q 2002, and they had better get close to meeting that deadline. If they do, then Palomino has a very short lifespan as king of the hill, so short as to make you wonder a little whether AMD might not be better off just skipping the whole thing on the desktop and just push Thoroughbred ahead. If they don’t, AMD is in trouble.
It’s pretty funny that at least one of the people who said I knew nothing about business showed no concept of amortization, which is a pretty fundamental business concept.
Briefly, if you spend money on a capital asset like production equipment, accounting practices say you should not claim the entire cost of the asset right away, but rather over the useful life of the equipment. If you can’t use the equipment after a certain point in time (which certainly would be the case for CPU manufacturing equipment), you don’t “save” any money through amortization by delay.
If you spend $120 on a piece of equipment, and you use it a year, the amortization expense is $10 a month ($120/12). If you delay six months, and can only use it for six months, you don’t have any amortization expense for the first six months, but for the last six, it’s now $20 a month ($120/6).
Sure, you could keep Palominos in production an extra six months, but then what happens is you get hit from the revenue end.
Those Palominos are probably not going to make you much money, and you’d probably be better off converting to what will be the more lucrative .13 micron chips.
A few pointed out that AMD might save on interest costs. True, but the cost involved in what will be a limited retooling isn’t too big, so the interest on that for a few months isn’t much, and more importantly, if AMD had just discovered this principle
This New-Fangled Thing Called “Competition”
In 2002, we’re going to start seeing something most of us have never seen, a competitive top-to-bottom CPU market. For many years, of course, Intel ruled.
Then Intel fell asleep, and AMD has played (sort of) top dog. Some people seem to remain in a monopoly mindset and don’t realize the rules have changed. I had one person citing Intel history like it was Holy Writ and no one could possibly find another way to do it.
You don’t think the rules have changed? Ask anybody who just bought a 1.33Ghz Athlon processor for $250 or less. Is that playing by the old Intel pricing rules? Nope. Nor is it selling inferior products at a lower price; the 1.33Ghz is the best processor out there.
There’s more than one way to maximize profits. How much money can you make from an $800 processor when (almost) nobody buys it? You can’t make money from people just looking at it. You do make money when a lot of people spend an extra $50 or $100 than they would have otherwise to get the top-of-the-line.
If you graphed the number of processors Intel sells at various speeds, it probably would look like a pyramid, very few on top, most towards the bottom. AMD’s would look more like a box, with not that many fewer processors being sold at the top than towards the bottom.
Maybe half of you telling me about the new processors you just bought are buying 1333Mhz processors. You don’t want to play code hunting games just to save $50 at most, and you’re not crazy or foolish to do so when that is the cost difference. If that 1333Mhz processor cost $1,000 or $500, would you be so cavalier about the extra cash? Hell no, most of you would be scrounging around for codes, too.
This is a new world. The old one is gone, and good riddance.
Another Possible Reason Why AMD Might Be Doing What It’s Doing
You try to build high-speed Palomino. You have problems that delay you for a while. When working through the problems, you come up with a solution, maybe some of the earlier ones can be used on TBirds right away. The good news is that not only does the solution work, it will get you to a faster speed than you previously thought you could reach. The bad news is it’s going to take you six months to implement the latest solution.
AMD takes a look at the new improved TBirds. They look OK. They take a look at Intel; it’s still snoring. So they say OK.
Conspiracy Theorist To The Paranoid
One person called me a “conspiracy theorist,” like that answered anything.
Let’s face it, AMD is a pretty secretive organization. Wouldn’t be any need to read tea leaves if they just said what was going on.
But when you tell an investment seminar that you’re going into regular production of Palomino within a couple weeks, and within a week, you say “not for another six months,” that’s a pretty sudden turnabout, and “we just noticed we didn’t need to do it” is not too believable. Even if you did believe it, were these guys asleep beforehand? To me at least, the official explanation is less plausible than any I’ve come up with, and no defender of the faith has come up with anything that has substantially changed my mind.