So there is a way to calculate a theoretical limit of a given operation, given diminishing returns.

Right now hashing power is way too high, the next retarget is basically 8 days away (check out block explorer's eta function).

Retargets have been 50% for the last few bits.

We are at the 131,000ish block level right now. 210k - 131k = 79k. So 79k blocks to complete before only 25 btc's are rewarded instead of 50btc's.

79k blocks / 2016 = 39 difficulty increases.

Assume a single miner is rocking 1.0Ghash, and thus generating 1 BTC per day.

They will generate 14 coins over 14 days, and then there will be a retarget adding 50% difficulty. This will repeat every 2016 blocks until we hit the 210000th block, at which point generation is halved, so that will be my end-time.

So the limit, as x approaches 39, of 14 +14*(2/3)+14*(2/3)^2 +...+14*(2/3)^39 can be represented by a limit of some kind (math gurus jump in here):

I just used excel and calculated out the sums by dragging a simple formula for the coins per day, and each subsequent cell was 2/3 of the previous value.

I basically got 35 coins, maximum, for each 1.0Ghash by the 39th difficulty increase, assuming it actually does take 14 days per difficulty. If it takes less than 14 days, then obviously each 1.0Ghash is worth less than 35 bitcoins.

For me, at 330mhash, I mine 0.33 coins per day (4.6 coins in a 14 day difficulty right now). My maximum output is therefore ~11.5 coins, theoretically, over the next 39 difficulties. But really, once I get to the 6th difficulty, I'll have already created ~11 coins. So I will drop out in 6 difficulty increases, as I don't see much point in spending another 33 difficulties to get 1 coin.

So for those of you investing in mining rigs. I'd take a look at these #'s.