After 28 nm ships in quantity, and if they ever get 20 nm mining chips to work, that's when the REAL test will begin, of who is really committed to Bitcoin.
Because after that, there will be no more technological edge to push. The only way to increase your earnings will be to add hashpower, at whatever price it will cost (hint: 20nm circuit lithography machines are NOT cheap).
So, I think the weak hands and fly-by-night operations will be shaken out, and only the dedicated/properly planned mining operations will remain.
Network hash rate will become more stably predicted, and it will start to fall into a more linear range.
And, most importantly, I think BTC price will stabilize and will not have so many wild swings.
This will be good, in the long run. In the short run, people will whine
This sounds like : 640K is more memory than anyone will ever need on a computer.
Do you honestly think people will stop above 5mm ?
ps. Of course it's a myth that Bill Gates ever said that
I'm not saying that we won't get to less than 20nm circuits. I'm just saying it's pretty easy to go from 110 nm (1st gen ASICS), down to 65 nm, then 55 nm, then 28 nm, because that technology already exists and it's mature.
20 nm is on the horizon right now for later on this year. Beyond that, you have 12 nm tech, that they are working on for a couple of years down the road (and smaller).
But, your average fly-by-night BTC hardware company doubtfully has the deep pockets that is required for that kind of hardware investment, unless they are backed by HUGE amounts of venture capital. Not saying it cannot happen, it's just less likely to given current BTC prices.
With 10 times current BTC price, maybe.
Quantum Bitcoing mining, anyone?