In the coming months, anyone with 65nm BFL ASICs will start making less and less return for the same energy consumption, but avalons will hit that point first. Will be a similar case for anyone with the less efficient 28nm ASICs. KNC miners are pretty inefficient as far as 28nm process bitcoin ASICs are concerned, and they will be phased out long before BFL Monarch. This will give OGNasty more time to contemplate the 3rd / 4th generation bitcoin ASICs
Yup. Difficulty jumps are going to slow down because: 1) older ASICs are going to get shut down, 2) the transition to ASICs is largely over, 3) perpetual exponential growth is impossible, and 4) there is nothing "after ASICs," just better ASICs. From here on out it is going to be closer to Moore's Law, which is (very roughly) 100% per year not 100% per month.
Most of the ROI calculators assume a fixed percentage difficulty growth per month. That's wrong.
Keep an eye out because I think there are going to be some good deals on the secondary market when people do not understand this and are dumping their miners or their pre-orders.
I'm very optimistic about the future of mining, but it is a competitive market where success comes from being smart about it, not getting ripped off, buying the right products and operating efficiently.