Trade Shows: A Dying Phenomenon
Shows like Comdex are slowly but surely dying, and there’s a reason for that.
Trade shows need to meet two purposes: to tell its participants what is going on now and provide a platform to tell them (and the world) what is going to be going on.
The Internet has pretty much obliterated the reason for the first. One can generally find out a good deal more about products from one’s desktop, and can much more easily avoid or ignore the sales pitches.
For instance, Pioneer is demonstrating its dual 8X DVD burner. That’s nice, but once you say “we’re going to have a dual 8X DVD-burner,” anyone keeping up will know or can easily figure out what the ramifications of that will be. You don’t really need to watch one in action to believe them.
Then the question comes down to “when and how much.” In the case of Pioneer, it’s “maybe December, if the DVD Forum says OK” with a list price of $300, figure street price of $225-250.
No convention needed.
The second purpose of a trade show is to provide a media platform on which to provide new information not available on the Internet.
Unfortunately, that isn’t happening all too much, either. Rather, you get whatever the companies want to tell you, which is usually not what serious onlookers want to know to plan ahead.
Let’s see what will be important in the months ahead on the CPU front (which unfortunately, you’ll hear little about at the show).
Can You Get It Out The Door?
For Intel, the question is simple. “When can you get 90nm out the door, and what kind of shape will it be in when you do so?”
As you’ll see, the answer to this single question will have a huge impact on overclockers throughout 2004.
Intel has been as bad as AMD was with Hammer when it comes to this.
Remember that Prescott is almost a year late. It was supposed to be out last spring.
Since Intel isn’t talking, one has to try to piece together hints as to what the problem might be from what they are saying, and most evidence indicates that Intel has the same kind of problem that plagued AMD with Hammer. They can make them, but they can’t make them fast.
It’s not clear at all why they can’t make them fast; there’s a number of possible reasons and combos of that reason, but we just don’t know.
What we do know is that the overclocker market will be a lot different in 2004 if there’s a lot of Prescotts with even relatively modest overclocking potential floating around than if there aren’t.
The reason for that is not so much the Prescott platform per se, but rather the impact it will have on AMD and its pricing. If you have a marketplace next April or May where you’ll have $175 Prescotts able to run at 4GHz easily, most likely beating overclocked A64s and getting at least close to overclocked FXs, the overclocking market will be quite different than if this is not the case.
Should it be the case, most AMD enthusiasts will most likely sit on their wallets until AMD offers them as good or better a deal.
Should it not be the case, (if Prescotts don’t come out, or more likely, come out but aren’t very overclockable), there will be less competitive pressure on AMD, and higher AMD prices.
The viability of the Athlon 64 itself in the overclockers market depends on what happens with Intel. Bartons outsell PIVs in this market even though PIVs perform a bit better because between the CPU, the mobo and the memory, the core platform is $150-200 cheaper. What happens if the cost difference is less than $100, or there is no difference at all?
Looking at the FX, it’s at least 50-50 it will match or beat an overclocked Prescott platform. Would AMD enthusiasts pay more for an FX platform? How much more? Remember, for overclockers, it’s not a matter of laying $650 out for one as opposed to $750 for another. It becomes a matter of laying out $175 for one versus “now what do you have for me, AMD?”
2004 will be a year where the interests of the overclocker’s market will be quite different than that of the “regular” market. As we’ve just pointed out, Intel can be more expensive to the “regular” market, but be cheaper in our market.
It Takes More Than A Processor
AMD really needs to start filling in piece of the puzzle before they start cranking up production.
This is going to be more important than ever because AMD is giving every indication of trying to break out of the cheap home user and kiddie hot box market and find a new user base of, well, typical Intel buyers.
Not that there’s anything wrong with the cheap home user and kiddie hot box market; I’m part of it. It’s AMD that is dissatisfied with it, probably because they haven’t been able to make money from it.
That’s fine. However, if that is what you want to do, you don’t do that just by jacking up prices and talking high-end. The market Intel has and AMD wants has a much different mindset, and different needs and desires than Joe Sixpack and his geeky kid. It demands more for their money, and what they want needs to be built into the company, not the CPU.
It’s not a matter of selling to richer geeks. It’s more like trying to sell to Mac users.
For instance, the item I’ve been looking for in Comdex reporting is socket 939 boards. So far, I’m not seeing them (though it is early).
AMD is going to need to have those available in order to be able to sell FXs in any appreciable number. This is a situation where absence means more than presence. If these boards aren’t around for Comdex, they’re probably not going to be around early next year, either.
But let’s look deeper into this matter. Why are we looking for socket 939 boards, and why might they be late? It’s because AMD didn’t decide to have a dual-channel memory architecture for regular desktops until the last second, months after Intel was producing such a product, and they’ve been scrambling to build one ever since.
That may not be too big a deal to the current AMD user base (they judge on different criteria). It is to the new markets AMD wants.
AMD also needs to make it clear just what processors are going to work with what memory, and soon. If they plan to do DDR now and DDR2 in late 2005, fine, but say so, real soon now. They need to also say what processors are going to work with which type memory and when, again, real soon now.
This may seem like an esoteric trivial item, and one that doesn’t matter much to the people who buy AMD today. However, it really is another indicator of whether AMD is going to be a serious contender for corporate business, or will continue to be taken as seriously as Bozo the Clown in corporate halls.
Your average teenage overclocker may not look down the road that far (though he should). Your average corporate buyer does, though. Why do you think Intel puts out news publicly about platforms well ahead of time? It’s because the corporate world wants it.
Items like benchmarks are secondary to this world. The corporate world wants its supplier to be predictable and reliable. It doesn’t like sudden changes of plan and being blind-sided. Intel is pretty good at the first. AMD unfortunately has a track record which is pretty good at the second.
This is the real, deep-down reason why AMD does so poorly in most of the upper echelon markets. This is why AMD gained so little the last time it had a technological lead over Intel (and one rather greater than anything we’re likely to see in 2004, and for a longer period of time).
This is the real, deep-down reason why Dell wants nothing to do with AMD, and why AMD has lost a number of big OEM accounts in the last few years. Yes, Intel pressures companies away from AMD, but this is the core reason why Intel has the power to do so. A lot of companies can’t afford to call Intel’s bluff and say, “You do that, we go all AMD.”
The problem AMD has is not the CPUs; it’s the company.
If AMD really wants to break out of the cheap home user and kiddie hot box market and start taking serious market share away from Intel, they have to deliver more than a good processor. They have to stop running the company by the seat of their pants and focus instead on being a good company in the ways Intel is.
They have to stop talking about being customer-centric and start being customer-centric the ways the customers they now want demand.