If mining isn't profitable anymore someone will stop mining, difficulty will decrease, and mining will be profitable again.
Bitcoin mining looks to be headed that way. Because of the lag between the decision to set up a big mining operation and it going into service, the big end of the mining industry can overshoot on capacity, so that everybody loses money. Eventually, some of the players stop mining, and the market does auto-adjust itself. The way this happens is called "bankruptcy".
ASICs can improve, but not a lot more; the limits of what fabs can build are being reached. Once that happens, Bitcoin mining performance will improve at roughly the same rate as CPU performance. (This is slower than Moore's law; flash memory improves with gate density, but compute-heavy chips are heat-removal limited. That's why CPUs have clocked at roughly 3GHz for a decade.)
There's a motivation for Bitcoin mining in China that some may not have realized. It's a way to turn yuan into dollars or euros. China's exchange controls make it difficult to convert yuan to dollars. The big Bitcoin run-up in China was because, for a few months, buying Bitcoin with yuan, sending the Bitcoins outside China, and selling them for dollars was cheap, safe, and legal. Once the People's Bank of China cracked down, it stopped being cheap, safe, and legal. (China residents can't just use their debit cards to buy Bitcoins on line any more.) Hence the Bitcoin crash.
Setting up a Bitcoin mining operation in China and exporting the Bitcoins is just fine, though. China encourages manufacturing and exporting. Buying equipment with yuan and getting paid for your products in dollars is perfectly normal. That's what a Bitcoin mining operation does. So, no trouble with the authorities.