Lol, the current difficulty is what, 8.9million? So at 91mill your 66GHs Avalon is going to do a coin every 3 days. So much for the ROI. At least they will be dirt cheap if they keep to their pricing formula. They will probably not get enough return on that formula to keep making them.
difficulty will go up about 10X by August... So if you have a second batch Avalon, there will have potentially a short time to make good money and then it may be a couple dollars a day. Depending on the BTC price.. So do you buy, sell or hold? Have an Avalon batch #2 and #3. Any feedback would be great (please don't pm me about selling them)!
So let's analyze your statement "difficulty will go up about 10X by August":
Let's make it by September because let's face it none of the players have met their self-made deadlines strictly so far.
Current Global Hashrate: 65 TH/s
Current Difficulty: 9.0 m
By
August September:
Avalon batch 2 + 3: 1200 X 65 GH/s = 78 TH/s
Avalon raw chip increase: 500k X 282 MH/s = 141 TH/s
Big Time Coin hashrate increase: 500 X 96 GH/s = 48 TH/s
BFL hashrate increase (wild guess) = 50 - 100 TH/s
ASICMINER hashrate increase = 50 - 200 TH/s
Total hashrate increase: 367 - 567 TH/s
Difficulty after adjustment in September:
(9/65) X (367) =
50.8 mto
(9/65) X (567) =
78.5 mOr an increase of 5.6 X current to 8.7 X current difficulty.
Also, people will switch off their GPU miners and FPGA miners around this time or soon thereafter, mitigating the difficulty increase for September - December. And there are no new technologies on the horizon that will exponentially increase difficulty thereafter, it will grow linearly. So while Avalon's difficulty calculation of doubling every month seems accurate for the next 3 months, it will not double any more after that and revert to more linear, less exponential growth.
So your Batch #3 avalon, assuming a cost of 100 Bitcoin and a hashrate of 96 GH/s, recieved in June or July, could potentially break even in 2 or 3 months. Definitely buy and hold, you will find no better ROI anywhere.