I read your story on AMD's server prospects. I agree with the point that it will certainly be an uphill struggle for AMD to gain market acceptance in the server market, but I think that you are forgetting about Sun. I believe that Sun had announced plans to use the Opteron in an upcoming line of Cobalt series servers. Sun is most definitely a tier-one server vendor. Also, I heard that Fujitsu-Siemens was planning to support the Opteron. I was surprised to read that Fujitsu had not made any comment in that regard. Maybe I had heard wrong but it may be worth checking into.
Like you indicated in you article, the other tier-one vendors (IBM, Dell,
HP/Compaq) will make an Opteron server only if there is a business case to do so. However, I do expect the Opteron to be the darling of the tier-two server vendors and the "do-it-yourself" market. I believe that the tier-two and DIY make up only about 15% of the total market but it's a toe-hold into the market nonetheless.
Server buyers are a conservative bunch. It's proven that the more expensive a computer is, the more conservative the purchasing decision becomes. AMD currently has no track record in this market and it will be a tough sell to convince these buyers to go with unproven technology. It will take time for AMD to develop this track record--if they can manage to deliver consistency, high quality and value over the long haul. All the names in the server market including Intel were unknown and had to prove themselves at some point, so it's not impossible to conceive AMD achieving that stature someday.
If AMD can convince buyers in the server market that they can have at least Itanium performance or better at a Xeon price and they don't have to port their old applications to the new system, AMD may have a selling point. A lot of companies use custom-written applications that have been in service a long time and they really don't want to have them re-written to run on their new Itanium servers. They may go for the Opteron instead. It may become a compelling reason for some companies to buy AMD.
My prediction is that AMD will gain 15 to 20 percent of the server market by this time two years from now. It won't make them a world power in the server world, but they will be nothing to laugh at either. I read that AMD currently has 6 percent of the server market with the Athlon MP (and zero percent the year before). As you wrote, they have nothing to lose and everything to gain. I admire AMD for their lofty goals and I wish them well.
I will make another prediction. I think that the Opteron (ClawHammer) will be a very strong product in the desktop and workstation markets. In fact, I think that it will crush any hope that Intel has for the Itanium ever scaling down to the desktop or workstation market. I believe that the Itanium is doomed to be predominantly a "big-iron" server and specialty-market processor for its entire market life. Intel may be many things but they are not a stupid company. They will have to deal with the AMD competitive threat and release a 64-bit x86 chip of their own. The rumor is that they are already planning x86-64 extensions in their upcoming "Prescott" processor. Itanium will have to compete against both AMD's and Intel's own x86-64 bit processor. Currently the computer world revolves around Windows running on an Intel-based architecture (Wintel). x86-64 is a logical upgrade path to the 64-bit world for a Wintel user; the Itanium is not.
Cheers, thanks for writing such an interesting article.
Ed Gallagher
Information Technology Services
