That would be DRAM memory which is using the EUV. not V-NAND. So a problem for CXMT and Fujian Jinhua not YMTC.
Chinese semiconductor industry
- Thread starter Hendrik_2000
- Start date
- Status
If I remember correctly what I read here, the first SSMB underground facility will be ready for 2024/2025 and it will be mainly a scientific / technological project to make the SSMB concept to work in practice. From that you can expext at least some more years to make it viable in production, and so we are already toward 2030. I have no idea regarding plasma version, at the moment there is not even a prototype. Then there are tons of other issues like resists, masks, etc. that are still to be solved. Finally SMEE currently does not seem to be in mass production even with the non-immersion version of DUV and also its main suppliers are well behind (U-Precision will be ready in 2023, not clear about optical system, etc..).
Considering all the above, I can assume that by 2025, SMEE could be in mass production with immersion DUV (this is not a bad prediction, actually it is very optimistic one). And by the end of the decade toward 2028-2030 there will be a working prototype of EUVL machine, including needed localization of accessories like resist, masks, etc... Also this latter one is a very optimistic forecast, IMO.
Sorry to burst your bubbles, guys. 2024/2025 for a production worthy EUBL will not happen. It's not as trivial as you all think it is.
There is so much more than just EUV Source.
Let's be more realistic and not make unreasonable speculations like this.
Says who?
There is little evidence China needs "EUV" in the short term anyway. If the main intended application is AI then you are basically going to use it to replace human labor. You will end up having to pay people not to work, or they will end up in the middle of the street and eventually cause public disturbances. Sounds great for nations with a manpower shortage but China is not one of those. And a lot of the applications, I think, are kind of overrated. Let us say using AI to replace human drivers. This is supposed to be one of the main productivity benefits of next generation of AI technologies. Logistics and transport. In a country with a huge rail infrastructure, and which already has automated ports. You are going to automate the last mile only really. I see little evidence why you wouldn't do that even with 14nm or 7nm chips if you take some simplifying assumptions. i.e. you design the road for the vehicles, not design the vehicles for the roads. Which is what the US is trying to do. Which is orders of magnitude more complicated. With regards to use of AI in military applications, I think there are deep, deep moral issues in something like that. Having an AI which can pull the trigger is something I doubt a lot of people will stomach.
In short I think the best use case for AI will be as an enhanced assist for human operators. Not a total replacement which is what the industry is pushing for. And I doubt you will need EUV just to make enhanced assists in most cases.
China will likely need EUV in like a decade. But it will probably be able to have it by then. And by that time the industry will likely have hit a wall with lithography and either move to something else or stop evolving. I think we will just increasingly see more and more 3D structures in use and V-NAND is just the first.
In short I think the best use case for AI will be as an enhanced assist for human operators. Not a total replacement which is what the industry is pushing for. And I doubt you will need EUV just to make enhanced assists in most cases.
China will likely need EUV in like a decade. But it will probably be able to have it by then. And by that time the industry will likely have hit a wall with lithography and either move to something else or stop evolving. I think we will just increasingly see more and more 3D structures in use and V-NAND is just the first.
Last edited:
What are your sources for this and what is your expertise to assess anything as optimistic or pessimistic? In fact, what are anybody's sources for anything in this waste of space of a thread? The closest thing to an insider source we have states that China will have a first-generation EUV machine in 2025. That's pretty much the only fact that's been stated in nearly 1300 pages.
No, it does, and it's going to get it.
I see little evidence of this. Last time I heard China's EUV light source had, what was it, 50 Watt performance? Industrial machines are using 200 Watt or more by now. And that is just the light source. Where is the rest of the machine? Or the materials. Who is working on the materials? The masks and things like that? The Huawei sanctions started like 3 years ago and we still don't have immersion lithography. And you expect EUV in 5 years?
The AI race is a very special one because is the only advanced technology where US and China started more or less from the same point, at the same time. The gap between US and China in AI, or is not existent or is something around months / couple years maximum. This is in contrast with all other advanced technologies where China has been forced to close gaps of many years or even decades, always playing catch up from behind. Many have been closed, other are getting closed. But AI is unique, because, the discipline started at a point in history when both US and China had the required technical and scientific knowledge to develop this new field on a more or less equal footprint.
It would be highly unwise for China to give away this unique and precious opportunity and let the US to advance alone.
Yeah, last time you heard. Does the Chinese lithography industry give you regular updates?
Who says?
Not 5 years, 3. 2025 - 2022 = 3.
No way. That is not how the industry operates. To have it in mass production in 3 years it would have to be in test fabs producing chips at a low rate RIGHT NOW. So they fix the bugs and get the line and process to have low defect rate. So where is it? Where is the EUV test line and facility?
- Status