Because, as I said, we are at a special time in history. Today ASICs are new. in 5 years time they'll be old. Today, the state of the art advances so rapidly that current designs become obsolete quickly, rendering investment in them moot. That's partly because their technology is lagging behind the mainstream, and their rate of progression is largely how quickly they can adopt technology from the mainstream. In 5 years they'll have caught up with the mainstream, and then they'll advance at the stately pace of Moore's Law, which itself will slow down. It's partly because they are so new, and the benefit of ASICs is so large, that any old design yields big improvements, and then finding a better design is easier. There's lots of low-hanging fruit, relatively easy to pluck. After 5 years of this, finding new improvements will be hard because all the easy ones will already have been found. You'll be able to licence a proven design that any chip-maker could produce. They'll become more like commodities. Having your own ASIC will be like having your own hard disk drive. Sure, cloud storage exists, but it doesn't use some magic technology that isn't available to the general public.

I'm not talking about today. I'm trying to predict conditions in 5 years. I might have that wrong, but clearly, today mining is in a state of flux which will not be typical.