FYI, I used the same link to do my estimations and found anything below 400GH/s won't break even in time or at all. Taking 50 GH/s for example never breaks even or make a profit

With the estimated difficulty rising from 54mil to almost 67mil in the last week I seriously doubt any of this would be viable anymore

So by how much can a 50GH/s unit be overclocked for?

With the difficulty expected to be around 200mil by the time you guys ship; the only way you'd be able to break even on a unit is if you buy the 400GH/s one. Unfortunately I don't see the price for a BTC to go up any time soon to compensate for the slower rate due to difficulty. Unfortunately I don't have the money for a 400GH/s unit

Thanks for the feedback so far guys. I think I have my answer, will cancel my order and rather push my money into a different venture

My order is for Bitfury ASIC unit and costs 20 BTC. Delivery date has been confirmed for late October but may only happen in November.

I should be seeing 1 watt per GH in terms of power usage.

I might have been incorrect in terms of the increase in difficulty, what I meant was it would see a 50% increase every 2 weeks (or sooner) according to some models I've seen online. The only way I'll be seeing a ROI within 3 months after November is if I buy a 400GH/s unit, which is 120 BTC.

I'm not taking into account the cost of a BTC in $ by that time. I'm just wondering if I should rather take my money elsewhere and cancel my order

I recently ordered a 50GH/s unit that I would have (optimistically) wanted it to pay for itself in 3 months (delivery date November). It would appear now, with projections and the almost 50% increase in difficulty recently, that the trend for a 50%+ increase will continue. Especially with the time solving/confirming dropping from 10min to 5min.

It would appear that I won't ever even break even with this unit that I pre-ordered.

How accurate are the predictions that the difficulty will be at 178mil by November 2013?