I'm not sure anybody can compete with 28nm SOI at the moment (as in the first linked thread), but ASICMINER can work together with them!
Agree, in multiple levels (assuming these 28nm guys are legit, and their product is superior):
1) Buying a percentage of their production, would secure our share of the network
2) Making boards our selves would reduce the cost, hedging against those who would buy them assembled (=other miners have to pay profits of the assembly company)
3) Asicminer would be able to sell those assembled boards. Those would secure some of our profit / hashrate out there.
4) Asicminer should sell as much as possible those boards to shareholders, thus generating extra value to them (I'm greedy, I know)
5) Reduced R&D costs
What is stopping ASICminer from ordering 28nm chips ?
1) R&D costs
2) R&D time
2) Risk of failure, and loss of capital, and loss of hashrate due delay
3) higher chip price (if the guys are so large as the given impression)
I know that friedcat&co are very capable people, but why take risks when one can be without? At least we should diversify (some chips in house, some out sourced). And there is a reason why, Apple for example, uses Intel chips nowadays. We don't have to order them right away, there is 50TH+ still coming. But they are claiming that they will put out 1 PH, with superior hash/joule in the first run. And that would drop our share of the network to around 6% with the 12+50TH, and then there is other competition too. Buying for example 10% of their production would be beneficial, even if the chips never went online.
I meant that they could buy the chips from this other company if they are legit.