Difficulty is rising, but BTC price is rising even faster. For some rough estimates, difficulty was between 75,000 and 100,000 when BTC was between $0.75 and $1.00. Now BTC is over $3.00, (projected) difficulty spiked and crossed 150,000 but has leveled off a bit over the past 48 hours. It will probably spike again after new miners' equipment deliveries arrive. If BTC stays over $3.00, difficulty will still has to go over 300,000 before it is back on track with the previous ratio (it may happen in a month or two).
Furthermore, difficulty growth has been declining over the long term (
http://bitcoin.sipa.be/growth.png). This could be signifying the end of already bought video cards coming online. So new investments in mining are less and less competing with hardcore gamers who already have the equipment to start mining. In that case, new investments in mining will be more profitable than before.
But let's examine the possible risks that they'll be less profitable.
Previously, small investments in mining probably have been less profitable than small buys. The main reason for this is the competition with hardcore gamers. But as this demographic is exhausted, a new one could fill the gap.
The new demographic is Big Players (tens of thousands $$ and over). Some will choose not to buy much because it would push up the price (even in dark pools). Instead, they will invest in mining capacity. We would see signs of this happening if we see the difficulty factor outpacing the price (the opposite of what has been happening recently).
Evaluating the equipment options available to potential big players, they don't seem likely to be able to squeeze out the small miners in the short term. The estimate of the FPGA performance thread shows that although FPGAs blow away commodity AMDs on a hash per watt basis by at least some 5x-10x, they are probably more expensive by at least 3x. If the cost of electricity on a monthly basis for an AMD is some 5-10% of the cost of the card C, then over a number of months T, the total cost would be [C + (0.1C)T]. If the cost of the FPGA is 3C is then the FPGA beats the AMD after 20-40 months.
ArtForz has come out publicly as one such player. He claims in the IRC logs to have invested some $65k in 100 structured ASICs (not named as such explicitly, but "like FPGAs without the FP"). So this is a longer term bet that they will outperform commodity AMDs over the next 2-3 years. After that, an evaluation will have to be done again between next-gen structured ASICs or FPGAs, and next-gen AMDs.
The point is that signs of any such large investments (FPGA/structured ASIC as well as commodity AMD farms) by bigger players versus direct investments buying coin will be reflected in the difficulty factor outpacing the price of BTC. Right now, the signs are the opposite (BTC price outpacing the difficulty factor).
I'm a trader, not a miner, but even so my opinion is that a play in mining (big and small alike) is even smarter today, than a play in trading. The longer the play the more competitive they are. The trick to profiting in either is the same: selling high.
