I think this needs to be looked at as much from the expenditure side as the forecast fabrication side.

In essence, a years worth of investment is coming onto the network now and over the next few months.  A lot of new miners (like me) have spent their available cash on pre-orders.  I'm not sure that the market will be there to pay for the production of the ~ 6 PH/s that's been announced by the new ASIC foundries over the past couple of weeks.

Of course, I could be very wrong, and new corporate-scale mining companies with new external capital might appear, but the existing contract 'cloud' hashing services don't seem to me to be capturing that much interest.

So another way to look at the question would be to try and calculate retrospective month-by-month dollar spends, and projecting that curve and factoring in some approximate hashrate-per-dollar forecasts.