So far, the Chinese have been evolutionary in their approach. This is probably because they haven't had the capital, or expectations of reasonable returns, to invest in large die 20nm ASIC development. They deploy 'old technology' at scale, with tuning to meet short term market demands. After this current generation of Chinese mining technology, they will be faced with the choice of either having to make significant investment in their own 20nm ASICs or exit the market. My guess is that several Chinese manufacturers will join together in a consortium to fund the needed investment and share the results. Eventually, they will consolidate.
I agree that the chinese have been revolutionary but it is not because they lacked capital (hence no preorders). They did not go with 20nm for the same reason everyone else didn't, which is because the costs outweigh the benefits.
KNC was fortunate in their timing and invested profit from their successful first products and their mining revenue into large die 20nm technology. We'll soon discover what kind of performance this yields. My guess is a 1st in Class machine that they can then node shrink to 14nm without a lot of risk or expense to enable them to stay competitive well into the point on the mining profitability curve where others can not compete because there is too much risk to capital returns.
KNC 20nm will be ~0.7w/gh at the wall assuming the 30% increased efficiency per die shrink. This puts their 20nm chip on par with bitfuries 55nm chip in terms of efficiency. If bitfury makes a 28nm chip it will certainly destroy everything on the market today.
Also large die 20nm technology is not the ideal solution for bitcoin mining. Each chip will need a powerful cooler (most likely water cooled). A bunch of small and efficient 8mm chips with a shared heatsink is a much more cost effective solution.
Other existing mining equipment manufacturers fall somewhere between these two cases, some relying on die shrinks of old technology to stay competitive, others investing in 28nm and 40nm designs and mostly falling on their faces leaving a trail of unhappy pre-order investors in their wake.
The face falling has to do with the preorders not the process size.
So, the long and short of it is that we are getting to the point on the profitability curve where if a company hasn't already made the very large investments needed to stay competitive over the next few years, they are out of business. Its is clear to me that KNC, so far, has made all the correct strategic moves to ensure that they are a leader in this market for the foreseeable future.
Completely disagree. I think they are shooting themselves in the foot by going with a die shrink to 20nm instead of improving their design.