It depends on the electricity cost, the block-reward, the exchange rate, the difficulty, and the quality "Q" of the machine (number of hashes per joule). One interesting event that will occur in 2016 is the block-reward halving, which is discussed below. The upshot is that I don't think the block-reward halving should discourage you from mining as long as your electricity costs are less than about 9 cents per kW-hour.
(In the following, I'm assuming the exchange rate is $440 per bitcoin and the difficulty is fixed at 93448670796.3238.)
A simple (perhaps naive?) calculation implies that a good miner with "Q" ~ 2GH/Joule should be profitable in the long run at current exchange rates as long as your electricity cost is less than about 19 cents per kiloWatt hour.
With all the same parameters, but at half the block reward, you will still be profitable in the long run as long as electricity cost is less than about 9 or 10 cents per kiloWatt hour.
If you believe the advertised "Q" value of ~4Gh/joule for the S7 then even with the block reward halving you may still be profitable even at higher electricity cost... however once the S7 is available to you it is available to everyone and the difficulty will increase as a result. In general the difficulty always tends to increase, so be careful...
All the calculations presented above were performed at fixed difficulty and exchange rate.