We speculate that AMD’s move to take an “architectural” ARM license just a month ago (meaning product hits the market in about 2 years from now) may have been a tactical mistake and perhaps a missed window of opportunity.
I think AMD were right not to invest too much in ARM till about now. The ones who have gone into 64 bit ARM so far have real uses for it without having to attack the server market. There's no chance of appreciable general server sales till 2016 at the earliest, people would just be testing things before then. AMD wouldn't have any special use for the cores itself before then. The big problem ARM have if anything is nobody wanting to be first to market as much of the effort for that will not have a payback - better to wait and put out a better product when there's more of a market.
Market Data 
Markets 
AI
