I think Starcraft alluded to something important; memory doesn't really matter to a buyer. They buy the computer/platform and whatever memory it uses, they buy. While memory is part of the overall performance and price of the platform, virtually no one buys a platform because of the memory it uses. You do not see, for example, people talking about about their DDR2 system, or start with memory and pick the parts around it. They talk about their Athlon 64 or Pentium 4 system, and generally start with the processor and work down from there.

AMD is not significant enough to pull the market one way or another. They still struggle with around 16% share, and less in the high end, particularly servers. So, while they will help the trend, it will move towards DDR2 with or without them. Intel will control that.

Also keep in mind that there is always a legacy market. People are still buying RDRAM, and Samsung is still making it, for example. It is very, very expensive (I found myself in need of it when I bought a brand new OR840 motherboard on eBay. Easily twice the price of DDR). In fact you can still buy brand new EDO memory. SDRAM is very easy to get. DDR will be around for a long, long time to come, particularly since DDR2 is a minor improvement, if that. But, the real reason is the number of machines that people have bought that use DDR. These machines will be around for a while, and as long as they are around DDR will sell.

Their platforms not delivering is a matter of opinion, there are plenty of people happy with them and new technologies being introduced with them can more easily be seen as positive than negative. Some people are not going to be happy because they can't move their AGP card to a PCI-Express motherboard, but in terms of percentages, they are very low compared to people that simply buy a machine without moving lots of parts from another. They will be happy Intel included newer technologies since it makes the upgrade path that much better, rather than shackling them with AGP, which while adequate, will not be supported for as long and therefore have a shorter upgrade path. I think Intel's platform problems start and end with the P7. It sucks, and when the processor sucks there isn't all the much that will mitigate it.

You brought some new aspects on the table. I agree with you most of the time, but not everywhere.

First I'd like to ask a question: Would you really want to buy a current Intel P4 system? I wouldn't do that in 99% of the cases (the remaining 1% is the unlikely case where the Intel system is significantly cheaper at the same performance point or when I would be able to get the Intel system for free). As the situation is currently, I wouldn't say this is only a matter of preference. Enough reviews show that the P4 is outperformed by the A64, although the latter has lower heat dissipation and also on a lower price point. If you don't care for performance, thermals and power consumption, you'll be fine with Intel. Otherwise not, I'd say. But that might be a matter of preference if you don't care.

Then I really agree with you about the existing systems driving demand. But additionally I believe that OEMs and some buyers preferred intel systems with ddr memory over ddr2 for pricing reasons.

btw, AMD is growing very strong in the server market. Plus there is a huge installed base of ddr-memory in this sector (regardless of intel or amd), as upgrade cycles are a bit longer.

But primarily I posted the first comment to point out that you can't suggest "ddr2 is good to go" (anna did that in her article) without specifying who would actually need it. Owners of AMD systems do not (yet).