An expert can correct me if I am wrong, but what you say about "no buildup of advanced node capacity" is absolutely not true if 7nm chips are considered "advanced". SMIC is scaling that up as demand for the new Huawei phones vastly exceeds chip supply and production capacity.
Chinese semiconductor industry
- Thread starter Hendrik_2000
- Start date
- Status
i m not an expert but 7nm do count in advanced nodes. and this is true as well, SMIC's most advance Shanghai line is currently expanding. but he meant to say, not as expanding like mature nodes in mainland.
i believe the scale of advance node capacity will further increase from 2025 onward when domestic tools/lithography finish expansion and verification of tools.
I think he was pretty clear cut in saying there is NO buildup of advanced node capacity. That's quite an absolute assertion that cannot be accepted as truth without significant evidence. Especially since China was supposed to have no advanced node capacity to begin with, as per imperialist sanctions. It seems loopholes and shell companies/intermediaries still exist to get equipment where it needs to go.
He did not say advanced node capacity buildup is slower or lagging behind mature mode capacity buildup. Just reference the quote below. What he said was very absolute and straightforward.
Last edited:
Hmm, they need both. You are completely overstating this matured node fab build up.
If it actually meets China's demand, then why is doesn't mcu makers still come close to meet even 50% market share?
The 40/55nm capacity we see them add is a fraction of what's needed. It's a problem that tsmc is allowed to expand on mainland and take market share away.
Their own fault. I believe they sourced their non-Hisilicon components from Western companies only before the sanctions.
Does anyone know why Micron introdced EUV 2 Generations after SK/Samsung? It seems a bit strange that they onyl started using EUV now, when everyone else did it so nuch earlier.
I would imagine the tech between different types of DRAM to be similar. So if they can get to competitive IO speed in LPDDR5X, then maybe they can get to something competitive in DDR5 (assuming they get there). Which would be hugely importantly for AI and data center and such.
Of course in terms of phones/tablet/laptops, if they can get to 15nm process with LPDDR5x by 2025, then that will be huge for revenue
Still biggest news of the week imo
Alright, so I read through the SMEC Investor Q&A from November. A lot of interesting notes in there
Looks like they also fab RF switches & other consumer electronics stuff
SMEC pretty important even if it doesn't fab advanced nodes
SiC production up to 4k wpm by end of Q3 from 2k just 3 months earlier. 90% of that is used for main electric drive inverter. Looks like SMEC will get really good here. And they continue to increase 12-inch & in-inch wafers too
Huge gain here in MEMS INS & Lidar system. I think MEMS is one of the most critical semiconductor field that just don't get the attention. So many important tools need them
Claims its 8-inch RF filter tech is leading in China and used in 4G/5G areas. I think Newsonic & Huntersun use SMEC
Looks like they also fab RF switches & other consumer electronics stuff
Again, they claim here that their BAW filter tech is already in mass production and comparable to foreign IDMs, especially 5G filter, accepted by mobile end users. I don't know if this includes N77/N79
Look at this 80k wpm of 8-inch wafer IGBT production. Largest IGBT producer in China. Huge here in their domestication efforts
SMEC pretty important even if it doesn't fab advanced nodes
Interesting, this FT article had stuff that's not awful (I mean if you can ignore the usual think tank nonsense)
this is quite interesting. Now, I don't think 34 lithography steps vs 9 steps means 4x the time for the entire process, but just the multiple exposures here is quite interesting
so they said early days it was already at 30-40% in the pre-MP phase. That seems reasonable. I'm sure it's a lot higher now
so this is the interesting part. using 0.2 to 0.25 D0, I rounded yield rate on Ascend 910B to 25%. But using that D0, I got 70 to 75% for Kirin 9000S. Again, the yield here is close to what I think it is.
- Status