I know all this is pretty vague, but as I said, there are too many unknowns at this point for a precise profitability calculator to have any meaning. Also, it looks like these units will come out around the time that the reward halves, making it even more difficult. Those who like more transparency in their investments should stay away.
"Prediction is very difficult, especially about the future." - Niels Bohr
Also, consider that when you place the order today, you are spending your bitcoins at today's exchange rate.
If you were to, say, wait until October to place your order, the exchange rate could be higher (or lower, but let's say higher ... say $8).
So the person who pays for the $1,299 SC Single today at $6.65 pays about 195 BTC. The person who pays in October at a (hypothetical) pays about 162 BTC for the same product instead. But person who waits until BFL has the units "in the wild" didn't need to worry that BFL really could deliver on these units.
Yeah, that's something to consider. You could then buy 4 for the price of 3 at todays prices. So, in that scenario, there is effectively a 30% inrease in hash rate per dollar. But this might be balanced by the first month being most profitable. Furthermore, ordering now and get the hardware to mine before this hypothetical $6 to $8 increase. So, we need to consider the potential factor that you will have likely mined significant amount of coins before they become lower in supply (i.e. the 50 to 25 bitcoin discovery size). So, who knows.
And also consider that BFL will have to either charge less per hash rate once the difficultly goes too high. Afterall, when network hashing goes up 30x, who would want to buy at tody's prices? I suppose there will be some kind of market equilibrium...perhaps eventually. Miners won't invest if there is no reasonable time to ROI, and difficulty won't increase if miners don't invest.