"However, development of such process nodes and validation/qualification by customers will take some time,<...> as a result, Applied expects that companies in China with 14nm designs will continue to use that node and the services that the foundries can provide. Meanwhile, a hypothetical 17nm node (assuming it makes financial sense) could be used for an entirely different design. "
Will never happen. If someone in China made 17nm FinFET then the US would just sanction them again. Originally the ban was going to be on 14nm so that SMIC FinFET would get banned. But they made the rule 16nm specifically so that TSMC cannot increase its FinFET fab capacity in China either.
A process which might show up at SMIC, I think, is 22nm planar. HLMC, TSMC, and UMC have it. I think it has the same geometry as 28nm. So you can easily shrink a 28nm design to the 22nm process. Unlike FinFET it does not need a substantial redesign. It has 10% lower area, 15% better performance or 35% lower power consumption, than 28nm. Plus it probably won't be sanctioned.
Unfortunately SMIC's leading process designer is against developing more lower end nodes. It is not that I can blame him, since most people who had 20nm designs switched to FinFET and there is still a lot of volume in 28nm while 20nm production at TSMC is basically dead. But unlike 20nm, the 22nm process should be cheaper, since it requires less complex design rules.
It boggles my mind why the Japanese have not made the effort to ensure that the domestic market for more advanced IC chip nodes is not dominated by Japanese fabs. I have read from many sources here and elsewhere that t most advanced logic chips manufactured in Japan are at just 40 nm.
The Japanese stopped making advanced logic fabs after the last PlayStation 3 came out. The PlayStation 4 and later basically have X86 processors. As for making new logic fabs, for that you need to have enough volume to use them to capacity.
