The foundries are currently ramping up their 1st Generation 7nm Production and it looks from an article just put out by Scotten Jones that most of the volume targeting the 7nm process node is waiting for EUV in 2019. Check slide 13.
Will manfacturers be able to continue to get better efficiency from Bitcoin miners by moving to these more advanced nodes? My first inclination is to say yes. Moving to 7nm is going to help achieve better efficiency.
The glitch to this thesis that I am trying to wrap my head around is The E-bit E10 18T. They say they are using 10nm Samsung chips but are only able to get 10% better efficiency than the 16nm TSMC Antminer S9. On TSMC website they say that moving from 16nm to 10nm gives ~35% better efficiency.
http://www.tsmc.com/english/dedicatedFoundry/technology/10nm.htmWhy didn't E-bit achieve better efficiency than that what they are announcing? Do you think they compromised efficiency to increase speed to shave some money on silicon? Or is the Samsung 10nm that overrated?
Their cost is also another issue. With E-Bit E10 18T miners for March release sold out at price near $5000.
https://www.semiwiki.com/forum/files/Scott%20Jones%20ISS%202018%20Presentation.pdf