wbmw, you make a sensible argument but there is a weakness in it. While Intel improves IPF the rest of the market is not standing still. Sun, for instance, is developing a whole new architecture based on a massive multi-core approach. Will it work? Don't know yet. IBM is aggressively developing new Power versions.
The other question, still pondering, is this: Your list of companies - SGI, NEC, Unisys, FSC, Hitachi, and Bull - how much revenue did they see from IPF systems in 2004? How much can they (realistically) expect in 2005? How does this compare with the size of the overall market they are addressing?
Also, you used the phrase an integral part of all their server lines. I am not sure what that means. I know that SGI is betting their company on IPF, but SGI is small potatoes. The others have several irons in the fire, correct? Only HP and SGI have burned their bridges, to my knowlege.
However I do find your conclusion to be pretty realistic (which probably means not too different from my own). You post:
Having said all that, I see IPF filling up its area of the market - roughly 30%, right? - as soon as HP transitions customers over completely, which will happen over the next year or two.
I believe that would still leave the majority of IPF sales from HP. Our area of disagreement is limited to the 10 to 15% of the market outside HP which you expect to be met by IPF. I think that number will be in the low single digits, making for an overall IPF penetration of about 20% of the target market.
Anyhow, you answered the item I just posted to Doug - are there any other rational posters out there who can talk about IPF? I am glad the answer is yes!
Market Data 
Markets 
AI
