The idea that we will see a 50% jump in difficulty every time for the next 39 adjustments is VERY unreasonable. Let's do the math here.. 1.5^39=7371554 and change. Basically you are assuming that the hashing power of the network will increase by ~7.5 million times it's current value. At the systems current level of maturity I seriously doubt that.

Let's do some more estimates around this figure.. Assuming 15000 GPUs make up todays network. (I'm probably off by quite a bit, but not by any significant order of magnitude) For your assumption to hold true, in about 2 years we would be having 15000*7371554=110573310000 GPUs in action. Or about 20 GPUs per person on the planet. I'm not even sure if we have enough power on the planet to run that many cards.

If you truly beleive this then I suggest you go out and put all your money in AMD stock on whatever exchange they are being traded. For everyone else I suggest you just keep your miners running. Growth in computing power/increases in difficulty will at the end of the day be connected to the profitability of this kind of operation. Right now the profits from mining are insane, and we have lots of people moving in to take advantage of that fact. Eventually we will get to a point where you need a full year or so to break even, and at that point people will probably stop adding more power to the network. But as long as we are seeing crazy profits (like less than 3 months to completely pay off a rig) we will keep getting more and more rigs set up at the rate we are seeing now.