In 6-8 months a new technology will be invented to mine at huge speed, and probably will cost the same as today flag ship of (insert company here). Minimum speed should be alt least 10 ph/s to compensate difficulty
Probably not. 28nm is very likely as good as it gets for 2014. Even if a new 28nm chip was 2x more efficient than any other spec on paper it would still mean about 400W per TH, or 4 MW (as in a small power plant) for 10 PH/s. In 2015 we may see 20nm chips but the days of 1000x the peformance are over. 20nm chips is going to offer maybe 30% higher efficiency for the same cost, or 20% higher speed for the same cost.
Miners will get cheaper (maybe) and rigs will get larger but you really are limited to 2 TH/s on a "normal" US 120V outlet with current chips. Even a break through chip would be lucky to triple that. Prices may fall but they can only fall so far. At $0.30 per GH/s (90% cheaper than anything offered) a 10 PH/s rig would be ~$3M USD.