Mining will be almost impossible by the end of the year when difficulty is 100 billion. Companies are having trouble making newer hardware to offset the fast increase in difficulty.
Use it in your advantage. If by the end of the year you are running hardware that has already paid itself, and network difficulty freezes because returns are so low that nobody buy new miners (mining is a risky business, no person nor company buys hardware if they wont reach ROI within 12 months), then you got yourself a steady income until the next technology jump (in 2-3 years). If you multiply it by having not 1, but few miners, you got yourself a fat second salary.
There is nothing more efficient that 20nm. That is already the state of the art. Not even Intel has got smaller than that.
28/20 = 1.4. Neptunes should consume about 30%-40% less electricity than TerraminersIV/Jupiters , if both work at same clock speed. TerraMiners IV will get obsoleted long before Neptunes do.
That's at least my master plan. My Neptunes are fully paid. And I have another bit of coins waiting for buying few TH/s in KNC datacenter (coming second half 2014), however I am more confident in the Neptunes than in just buying hashing power.
By the end of the year difficulty will be 500-900 Billion. Put that in the BTC calculator for profit return. A 30TH unit will make $10 profit a day.