So that is $2/day/THS operating expense.
Up front cost looks like $895/THS. This will produce roughly 1.9 BTC by Sept 2016 and then essentially flatlines at 10% difficulty increases. It goes negative on operating expense after producing 1.4 BTC. This assumes BTC makes it back to $500. At $400 it goes negative on operating cost by Feb 2015 after producing 1.3 BTC.

At BTC less than $400 - you should spend the fiat on BTC rather than buying hashlets. In fact, if GAW was confident difficulty increases do not plummet, they can buy the BTC and pay it out over the productive life of the hashing power. They will pay out less BTC than the amount purchased and get $2/day/THS from the credulous customer.

I don't see any scenario in which one of these going into production today is better than just buying the BTC. BTC needs to recover to over $500 before the analysis even becomes plausible. I don't know what they are guaranteeing, but low BTC price today means people should buy BTC rather than buy new hashing power.

I think GAW promised they would never go negative on operating costs. So lets assume BTC eventually stays at $400 during the productive live of this hashing power. That means the hashing power produces that same 1.9 BTC before flatlining. You will have paid them $290 in operating cost. Today, if you can buy BTC @ $400, you get 2.2 guaranteed and no operating cost payout. Note, if BTC price increases during the productive life, you end up paying more in operating cost.

So lets assume average difficulty increase drops all the way to 8% (unlikely, but possible), then you end up paying $360 in operating costs before GAW zeroes them out but now will produce 2.3 BTC.

So right now $400 BTC price, you have to assume difficulty increases drop below 8% just to get more BTC from the mining than from  just buying BTC.

This is a great deal for GAW.