Actually, I'd argue that GPU mining mostly leveled off when positive ROI was around 8 months, then the BTC price drove it to break-even with electricity. I expect to see the same thing with ASICs - people will stop buying when it takes them more than 8 months to recover their initial investment. The caveat is that preorders screw up the estimates, since no one really knows how much is on preorder, and so no one can properly estimate their break even points. It could mean we accidentally shoot past the 8 month break-even window to something much longer.
Yeah, I used to think everyone would get scared off and not invest, keeping mining profitable. But at this point it looks like the market may end up completely flooded.
Right now though, a HashFast unit would break even in 10 days. At increased difficulty (quadrupled by BFL + this alone), you're still only looking at 40 days. Difficulty would have to rise to 24 times its current rate in order to see the 8 month break even point. I don't see that happening by October by any means. Because of this, I think this is still a reasonable investment, should it be legitimate and they deliver on time.
Right now we have about 350Th/s. Avalon is rumored to have 2 million chips on order, which is 700Th/s. This sale is 250Th total. Labcoin expects to have 500Th/s by november, as does BTCGarden. IceDrill plans to have 500Th/s on order from FastHash. Let's ignore BFL for now.
But you also have KnC. We actually don't know how many units they've sold, but if they've sold 2,500 units that's going to be about 1PH just by itself.
That comes out to
3.5 petahash online by NovemberWhich is when Cointerra starts to ship.
Now, maybe some of these companies won't ship. However, if you're going to assume HashFast ships it doesn't make much sense to assume no one else does. And I'm also ignoring ActiveMining and BFL.
So, if all that's online before you start, you might be looking at around $60/day, or $1800 a month, before further difficulty increases from newer, cheaper chips.
So yeah. I think it'll take a long, long time to ROI if all the other competitors execute.
.
