Huh, don't server cpus generally cost over $300 ?
Reducing your die size is not going to increase demand unless you cut the price in which case you have lost the die cost advantage but you may make up for it by selling more volume at the lower price at the same margin. I don't know if the Server market reacts that wildly to cpu price cuts of $20 ;-). I think the only benefits on the Server side are more performance and lower power as this market is not averse to paying any price if you give them competitive edge.
Yes, mobile is a given *if* it is lower power as you are entering new markets then.
The only reason to do it for A64 desktop first before Sempron is twofold. You may reduce price to spark more demand at higher margin than Sempron. Also *if* 90nm has an advantage over 130nm you want your higher price cpus to get it first otherwise Ed and the o/c'ers will just take advantage of the lower chip and ignore the higher priced ones. It used to be that every process shrink did give you an advantage but the jury is still out on 90nm ;-).
However if the price of your low-cost models drops below cost price than the decision is made for you as you can't really afford to give away *any* cpus for a loss ! It really is not as simple as the basic top-down method you stated which is the common mantra espoused these days. ;-)
News 
Market Data 
Discover 
AI
