Given that both IBM and Intel have already stated that Silicon has reached "the end of the road" at 7-10 nm (IBM's 7nm experimental work uses a HYBRID Silicon/Germanium wafer not the pure Silicon wafers most semiconductors have been made on for the last 30 years or so), it seems VERY likely that Moore's Law is finally about to break and it's going to be quite a bit longer between generations after folks get to 7nm.
might be interesting once the fabs move to a new process entirely (hybrid or whatever, ie not silicon based any more) as that will leave the current state of the art silicon fabs with lots of exess capacity. i cant see all those old tech fabs shutting down (or converting as that would mean basically all new equipment for the new process, so new fabs may be built instead, or existing ones being expanded, leave most old tech in place as its still useful) as lots of mainstream chips will still use them for run of the mill parts that dont need the extreme speed/power savings that the new process will offer. but those run of the mill chips wont take much capacity.
some asic makers will move to the new process but they will again be competing with the big boys and their cost will be substantial, amounting to what we more or less have now.. small runs, high price, low availability.. although they will be faster and more power efficient. so not a lot will be made at 1st.
so current fabs that have 90%+ of their capacity currently used by apple/amd/samsung/arm whatever will now be able to offer large runs of custom wafers at much cheaper prices. this may let asic makers (perhaps new ones as well at the few we have now) churn out "old tech" asic chips at much lower cost and much higher volume, as right now asic makers are competing for fab capacity. at the transition to a new process apple/samsung etc will move to those, leaving old fabs with excess capacity.
this may drive the prices of asics down to the point where asic miners are much more affordable and available. power efficiency will not be much better but the absurd cost of the chips will be lower, allowing new players into the the asic party.
this would mean
lots more asics on the network. difficulty would start to climb again just from the sheer numbers of new machines based on silicon, which should still be competitive if the price is right..
current farms are mainly limited by price per unit and power capacity. lower the price per unit and those who are not hugely power limited (small farms/home/hobby miners) can start to build out bigger farms. for example i have extra power capacity (125 spare amps with no upgrades) but cost of asics (mainly), plus somewhat limited availability prevents me from buying.
after all what keeps most asic miners in check besides power? price and availability. fix those and watch out.
just me guessing out loud, love to hear of any other thoughts on this.
