Great, we have another Xavier.
Chinese semiconductor industry
- Thread starter Hendrik_2000
- Start date
- Status
Let's not jump to conclusions, and even for Xavier I think his name has been dragged enough now, there's no need to beat that dead horse.
Robert's articles have all been a good source of information and often accurate, he deserves the opportunity to clarify what he means. I think he should be able to clarify why SMEE is not mentioned in this specific case.
There is no bloody way that he makes such a rookie mistake. Especially considering that he is in this very thread where SMEE has been mentioned in basically every other page.
As expected US is aiming at fully blocking DUV immersion. Note that this goes well beyond the current limits imposed on US firms, confirmed at just 16nm and below. But this does not come as a surprise: Qualcomm got permission to sell 4G smartphone chips to Huawei....but not Mediatek.
Now we will see what happens. For sure ASML will try to workaround this heavy limitation. But if they don't succeed and litho for 28nm is effectively banned, then 2024 will be the year SMEE goes in production with the 28nm.
I understand this statement is controversial. Many knowledgeable people like @hvpc probably do not agree, but I'm quite optimistic on this because:
1. Reliability: It is since November 2022 that a SMEE machine (not a Arfi one) is in debug/trial production at a fab. By Q1 2024 it will be 1.5 years of reliability improvements and software debugging. Reliability is a transversal property. When you reach reliability and robustness on a model, even if entry level, it means software and design is stable, and you can extend this to all your range. I say this based on my experience in machinery in general, not specifically in semiconductor machinery.
2. Performance: On January 2023 a 28nm machine was assembled at SMEE according to havoc. Plan was to delivery to customer in April 2023. By Q1 2024 it will be one full year of performance test at 28nm.
3. R&D: After a period of up and downs, now 130 Huawei engineers actively support SMEE that has 2000+ R&D people already. There are also some hundreds more at U-Precision. This is a huge talent firepower. By Q1 2024 is one year at full throttle speed with impressive amount of people thrown at the problem, and of course no money issues at all: litho machine is an absolute national priority.
4. ASML banned from 28nm: This is the icing on the cake. SMIC and HLMC both have 28nm expansion plans ongoing. If the ban for them is confirmed, then SMEE will fill the void. That's a given.
At the end of the day, in the medium/long term, to build China's 28nm foundry industry on a localized litho machine, so to achieve 100% localized equipment, it will be an important strategic achievement because 28nm is the bread and butter of a lot of industrial/consumer electronic and due to the wafer cost curve, it will remain like this for the foreseeable future.
China will build a totally independent and future-proof 28nm industrial base. This alone is worth the short-term pain.
Last edited:
A very good insight and summary, one can only look at Chinese reaction to see if this latest restriction may hamper them, there is NO hyperventilating like most of us here in SDF....lol It's very educating, I really like this back and forth discussion, we are all here to learn and boy I learned a lot.
Last edited:
There are subtle distinction to introduce in this discussion:
- Reproducibility of secret sauce (tech)
- I mean, secret sauce is theoretically reproducible, just "At what cost"-ing in terms of timeline, labor, materials, resources. The debate in realistic timelines expectation is a separate topic all together. Personally, it's just takes a lot of time, engineering is difficult...with that said, it's not dark arts or magic.
- Reproducibility of market conditions for success.
- What is difficult to reproducible is KLA's free market position (i.e., first mover advantage and early market entry leadership) and cumulative snowball effect of recycling profits into R&D leadership. The drivers of adoption in free market is profit maximization for shareholders, efficiency, and convenience. KLA is extremely difficult to overtake in normal economic condition.
But guess what? We no longer in Wonderland! Free market is over, we are at WAR now. The key drivers of adoption is no longer efficiency, convenience, rather national security. A new domestic "protected market" is a game changer to iterate product improvements over time, particularly with state subsidies. So hvpc insights are helpful but one needs to consider the broader political environment when discussing adoption factors, particularly in a war-time environment.
Last edited:
Robert writes market analysis articles. From the market perspective，SMEE's share in Chinese front end lithography machine is negligible，so in a sense he is not wrong
In Jan 2023，havok said that fully domestic 28nm production line will take 2-3 years to complete. See these posts below
Chinese semiconductor industry
I heard stories about companies wanting to buy SMEE frontend machines, my guess is a least their I-line and/or KrF machines but their delivery times were simple just too long, so they just defaulted to ASML or Canon instead. It doesn't matter if the immersion lithography machine can process, as...
www.sinodefenceforum.com
Chinese semiconductor industry
I heard stories about companies wanting to buy SMEE frontend machines, my guess is a least their I-line and/or KrF machines but their delivery times were simple just too long, so they just defaulted to ASML or Canon instead. It doesn't matter if the immersion lithography machine can process, as...
www.sinodefenceforum.com
So the timeline will be 2025-2026
No one here is saying that KLA equipment is not superior. What people are saying is that if KLA is not available China would use what is available and try to improve on that. They are not just gonna close the shop and go home because KLA is banned. TSMC is also more advanced compared to mainland fabs and that's where Huawei used to make their stuff. When they were banned from TSMC they didn't stop making chips.
It's not clear to me whether he is referring to a clean sheet fully domestic 28nm line being built from scratch, or whether he is referring to the 28nm capable DUV lithograph where it can slot into existing fabs or part of their expansion while other equipment can continue to use foreign products.
Or putting it another way, hypothetically speaking it is possible for the full tech stack for semiconductors at a given node to be indigenized, but still take time for a "fully domestic" line to take time to be completed due to the need to construct a new facility that is fully domestic overall, which is useful for bragging rights, but for China's semiconductor industry what's more important is whether the important lynchpins of the fabrication process can be indigenized and slotted into existing fabs or fabs that are being expanded.
.... putting it another way, if the question is "when will SSA800 immersion be ready for use in the fabrication process" -- that is a very different question to "when is a fully domestic 28nm production line going to be operational".
- Status