It makes sense for them to concentrate on that level for now. N+1 (aka 7 nm by techInsight) is likely the level that SMIC can get the best combination of yield and leading edge for the next 2 years. I think they can mass produce N+2, but it will be at lower yield and probably much higher cost.
Took at look at that article
It seems to me that no one is mentioning the interesting part about how Fujian Jinhua is suddenly back to life because it's getting Huawei orders. That means it must have procured the tools that it needs to produce DRAM for a while now and have already tested them out and started production. IIRC, it takes months to start production after all the tools of a production line moves into place. So, all of along the real problem facing them was not even lack of tools, but lack of clients after they got fully blacklisted.
The article itself also never bothered asking the question of where its suppliers are. Clearly if it wants to double production, it needs new tools. Who are these mysterious non-American suppliers? Would ASML sell to Fujian Jinhua with its reputation? Probably not. Maybe Nikon or Canon are willing to? My guess is that SMEE is possible here at least. Not sure what level of DRAM they are producing. Maybe not as advanced as CXMT. But if Jinhua can get every tool from non-American supplier to make DRAM, then I think maybe CXMT can get all the tools they need for their expansion? Maybe CXMT/YMTC won't be slowed down that much. What do you think @PopularScience ?
They also said this, but we know from Q3 earnings call that the bigger issue facing SMIC is US customers not wanting to continue business with them rather than not being able to purchase more Ammerican tools. These departures don't seem to be hurting them that much.
This can't be real. SMIC alone accounts for over 5% of global chip sales and they want to tell us that Chinese companies only account for 6.6% of their domestic market. That seems like complete nonsense.
So, this non-A lines are just referring to its current Beijing fab that is not designed to produce any advanced chips and the new SMIC Capital JV that is also only doing 28nm+ process. So yes, I'm sure the best they can do is 28nm at these lines. The lines were never meant to do better than that!
Again, SMIC is being extremely secretive about its SN1/SN2 plans. It's acting like expansions here aren't even happening despite ramped up purchases of ASML scanners that are clearly geared for SMSC JV. Recently, they changed the description of their business to 0.35微米到FinFET from 0.35 微米到 14 纳米. If they can't get all the ASML scanners they need, then their advanced node production will likely be slowed down. But I really don't think American tools are the limiting factor here. Will be interesting when we get the full years report and compare the revenue from SMSC JV vs the mid year report. If we see revenue more than double, then we know they are moving to more advanced nodes with higher production level.
SMIC is full of promises and confidence in its future while back at TSMC the future seems a little cloudy, last year is their full Zenith maybe in ten years time it will be part of Intel as one of its subsidiary.
