Just as a comparison, if you bought a Bitman S5 today as a newbie miner.
S5 is $413
PSU is $110
Raspberry Pi $56
Postage $23
Location USA, 10 cent per a watt.
Average difficulty 12 months 47 billion
BTC average price $225 over 12 months.
ROI is 14.28 months. Block Halving is in 18 months
Older miners, may have an existing PSU (11.28 months) or existing PC/Raspbery Pi (12.85 months) or both (10.28 months).
Currently, if you pop over to Havelock and buy some AMHASH1, on these identical variables you're looking at a 7-8 month ROI.
Units producing at 0.2watts per 1GHs are unlikely to out before June this year; you'd have made back most of your investment before difficulty rises towards 49.99 billion. Also, a lot of GPU manufacturers and CPU manufactuerers suffered 4-6 months delays when they went to 16nm fabrication (poor silicon yields equaled delays)
Additionally, (these are highly speculative estimates) because this is guessing at non-public information:
1.2watts per 1GHs 50PTHs is swtiched off, but still setup to be switiched on if BTC price rises.
1.2 watts per 1GHs 60PTHs is in locations were electricity is 9 cent per watt and switches off when difficulty passes 44 billion.
1watt per 1ghs will be switched off at a difficulty of 47.1 billion onwards. This is speculation, but about 160PTHs of this network hashpower is here.
0.7watt per 1ghs will be switched off at difficulty of 67.3 billions onwards. This stuff is quite new and there is about 80GHs of it on the network.
So, there are significant obstacles in place to halt rises in network difficulty
Great math and calculations but did you factor into that at these difficulties most cloud mining contracts and the average user will end however it does not mean the miners will be shutdown since their maintenance fees are usually much lower.