There is something I dont understand about this, maybe someone has a theory or can explain:
1.) Bitmain develops and builds CN ASIC.
Now:
2.a.) Bitmain (slowly or according to production) starts mining with them and difficulty rises.
or
2.b.) Bitmain just sells them and parallely mines with a certain amount because difficulty would increase to much?
In case 2.a. the miners go offline at bitmain and go later online at the customer. That should not increase difficulty.
In case 2.b. Bitmain would lose much profit. They could just mine as long as possible and keep the difficulty at an ASIC attractive profitable level and dont overproduce the ASICs.
So the problem in this case was that the devs forked away, right?
My question is, at what point do they decide to sell miners if there is no forking away?
They start selling when the market makes them sell. It's likely a combo of 2 markets, the mining landscape and 2. the hardware landscape. They can probably guess, based on difficulty, when competitors begin to "burn in" similar models. Really, though, they'd err on the side of selling too soon rather than selling too late. You have to understand that the first to market (and first to hash) with new generations of tech make the most money. Every wave that follows theoretically should make less money than the first because each subsequent wave waters down the market and increases the difficulty.
Additionally, it's important to understand that Bitmain and similar companies are not mining companies. They deal in hardware. As such, they are at the top of the food chain. They really should only mine if they can do so easily, for clear profit, and with little or zero expense. To the extent that their mining business ever impedes their hardware sales, they're moving themselves down a rung on the food chain and profits take a hit.