Rolling Stone said this about the recent DVD release of “Lord of the Rings: The Return of the King”:
“The only thing wrong with this bonus-packed two-disc presentation is that we all know a bigger DVD set, loaded with more extras and tons of deleted scenes, will be on its way before the end of the year. The intent is to make us pony up for the same movie twice. Millions fell for the scam with the first two Lord of the Rings DVDs, and millions will again.”
What does this have to do with AMD releasing a slower Hammer? It’s basically the same strategy: Buy a socket 754 system now, buy a socket 939 a little later, and then buy a socket 900 a little later than that.
A Strategy Cracking Or Just Phase II?
You might recall that AMD initially said that they would not release a desktop Hammer at speeds of less than 2GHz, and for a while, they did just that, configured Dresden for a million plus CPUs a quarter, and waited for the cash to roll in.
They then reversed their policy (but not too loudly) and came out with a 1.8GHz Athlon 64.
That helped a bit, but still wasn’t keeping Dresden busy, so now we have the 1.6GHz. If AMD keeps up its price equivalence with Intel, the official price will be $163, with retail prices being a bit lower than that.
It’s doubtful we’ll see any big sales breakthrough solely due to a $20 price cut.
If AMD had put this processor out last October or November or even January, it would have been terrific. Now, it’s too little, too late for anyone in this audience outside of those far, far behind the curve.
But in at least a few minds, everything is relative. We’ll see socket 939 very shortly, and even though a socket 939 Athlon 64 is the same basic processor as a socket 754, AMD will try to get a big premium out of that even though their competitor does no such thing in their product line.
The idea is to make socket 939 the primium brand for six months, and socket 754 the average person (or, far more importantly, OEM) will have to settle for, until they get rid of that and effectively toss socket 754 into the bargain heap.
Then it will be time to replace the socket 754 with a socket 939. Just like the Lord of the Rings DVD.
Or so the plan goes.
Artificial Scarcity Only Equals Lost Sales
People often ask me, “How is this any different than what Intel does?” There are some major differences.
AMD’s policy towards desktop Hammers has been one of creating artificial scarcities, apparently in the hope of shifting the perception of AMD buyers by buyers from that of a bargain CPU company to that of some premier CPU outlet.
Normally, CPU manufacturers start a new product line low. then work their way up the speed ramp, and as they do, the price of the first, entry processor drops.
AMD didn’t do that. Instead, they created an artificial “floor” rather high up in the food chain, above mainstream pricing. Then they slowly, quietly showed (or were forced to show due to lack of sales) that their “floor” had trap doors. Instead of solely building upward, they built up a little, then mostly (at least as measured by sales) built down.
The end result? Maybe they suckered a few extra people into buying early and paying more, but how many more people (and OEMs) over the last nine months ended up settling for an XP system when they rather would have rather paid a bit more for a Hammer system?
If AMD had been selling $150-175 lower-speed Hammers from the getgo, how many more of you would be owning one now? How much more money would AMD have made as a result?
Compare this to what Intel’s been doing lately. Northwoods initially started at 2GHz, which was low on the eventual ramp. To make OEMs happy, Intel rather quickly opened drilled even lower and introduced the 1.6A and 1.8A, which made both the OEMs and many of you very happy and got you buying systems sooner rather than later. Intel noticed that, and later on made sure that at least one processor in any new process line had a mainstream price (i.e., the 2.4C, and, for that matter, the 2.8 Prescott).
Doing this hardly hurt Intel’s profits. They’ve certainly sold many, many, many times more 1.6s and 1.8s and 2.4 and 2.6 and 2.8s than AMD has sold desktop Hammers (far more than the relative difference in marketshare).
Mind you, those 1.6s and 1.8s, etc. weren’t cheap, people generally paid from around $135-180 from them and OEMs probably not a whole lot less. They were, though, reasonably priced enough for lots of people to buy them.
Why can’t AMD do that?
Well, AMD tried a different road, and it essentially failed. So what are they going to do now?? Why, they’re going to do the same thing all over again, this time with socket 939. They’re going to offer only socket 939 processors high up in the food chain, and minimize their sales for what will be their mainstream product the rest of the year until they have to open up the trap doors again.
I carry on about this because based on what I see and what you tell me, all AMD is achieving by this is lost sales and less profitable sales. All their Hammer pricing policy has done has diverted sales to XPs, which is not the way it’s supposed to go.
The Sliding Slope
The next twelve to eighteen months are going to see differing platform after platform being tossed at us: socket 939 without PCI Express, socket 939 with PCI Express, 90nm Hammers with x86-64, 90nm Hammers without it, a brand new socket for DDR II, dual core configurations.
The only thing they have in common is that up to dual core, none of them will offer huge performance advantages for what will be mostly huge price tags.
Not surprisingly, most AMD fans are very, very inclined, especially after seeing the prices, to wait their “hero” out until the smoke starts clearing late in 2005 and Hammer systems start offering some hefty advantages over the good old socket A. This is not good.
What AMD Should Do
AMD ought to make socket 939 its mainstream solution now, and have a range of processors equivalent to whatever is available in socket 754. Now. Not next January. Now.
If they want to charge a few dozen dollars more for the socket 939 versions, that’s fine.
That way, people can buy knowing that they’ll get at least a good year of use out of anything they buy today, rather than waiting six months or more for the same opportunity, and getting less than a half-year’s use.
Give AMD fans the choice between socket 939 and 754. It’s better than making the average fan choose between socket 754 and socket A, which is actually what’s happening in the trenches.
One ought to keep in mind that the current head of AMD came from Motorola, which thoroughly followed the policy of premier pricing for its CPUs and pretty much premiered themselves out of the CPU market.
Will history repeat itself?