Have you thought about the amount of money needed for last year's network hashrate jump and the money needed for next year's jump? It's a big big difference. Jumping from 150TH/s to 120PH/s is not the same as a jump from 120PH to 3600PH/s. If we assume a 0.5$/GH * 3480PH= 1.74B$ needed for this diff jump. Don't really think that there are so many money that are just waiting to be spent on bitcoin hardware.
Difficulty only needs to increase fivefold before electricity costs eat up half your earnings. So you would only need to add ~500PH or $250M.
Common, who will manufacture miners that do 0.4-0.5 when the difficulty is over 300PH? Noone.
Unless someone makes a miner that does 0.2 (which is VERY not trivial and not many of current players can do that), and then his profit is just ~200$ a month from the start - why should he invest when ROI is 2 years?
After we each 200PH-300PH mark there is no reason to create more miners, and those who have miners will make good money for a long time.