When you plug your numbers into this calculator, do you ever stop to think about the practicality of 91B difficulty/650PH in a year? Is difficulty just a number that keeps going up inexplicably to you?
Ill grant you 650PH in a year is at the extreme upper end of what I think is even conceivable. But only a very small % of all bitcoins mined by the device will happen after the 6 month mark, which at 9B isnt exactly far fetched, so the last 6 months or so of that simulation are irrelevant compared to the first 6 months; just as 650PH at the end of 1 year isnt very likely, its equally unlikely to me the next 10 difficulty adjustments will all be smaller than the past 10. The next 10 will coincide with KnC B2+, Hashfast, ActM, Bitmine, Cointerra, Black arrow perhaps even BFL Monarch, next gen avalons and next gen asicminers hitting the market, the previous 10 were caused basically by just 2 vendors.
I agree that there is a lot of new hardware, but there is a practical cap on how much any one user is capable of running/any datacenter is capable of hosting. There exists some number x which is the number of people who still want to mine (vs speculate, which is becoming kind of awesome lately) and some number y which is the amount of power you're willing to contribute to the effort. Today, the network rate is x * y / average power efficiency. "New chips" change only the average power efficiency, which means that the network rate can only really change by a factor of the efficiency increase of 28nm chips to 65nm chips. Going from 3 PH to 650 PH means a 200 fold increase in GH/s/W, which is insane.
SO the simulation is very flawed (why doesnt anyone implement a calculator with a sigmoid function for difficulty?), the conclusion might still be true for the simple reason that I cant think of a mechanism that would cause asic manufacturing and deployment to stop as long as each of these miners is operationally reasonably profitable for those with access to cheap electricity. The real question therefore becomes how fast these vendors can produce and deploy, and KnC shipping up to 600 boxes per day gives us a glimpse of how fast this could go, particularly with 10 competing suppliers (at least one of which is using monster sized manufacturing facilities)..
I know it is the obligation of everyone in the forum to chase everyone off (moar BTC for me!) , so I know an optimistic outlook is a faux pas, but there is a cap on how many people want to buy ASICs and we're going to hit an inflection where hardware supply will simply exceed demand.