Chinese semiconductor industry

Status
Not open for further replies.

Hyper

Junior Member
Registered Member
Oh the U-Precision hater.
Yeah but still is in China, usually for the Chinese electronics market and still is a lot. Also most of the new capacity is local so that capacity will continue to increase.
You are unaware of the dangers. Them dominating production in China is worse than their production elsewhere. Once in China they will now actively suppress any viable competitor. They have to be stopped and let someone else succed.
 

AssassinsMace

Lieutenant General
I'm sure the Europeans are using US IP so it's a moot point on any collaboration with China because that's how the US has veto power in how it's used around the world hence how the US can tell a foreign company not to do business with China. The US wants the kind of control over others that when the US goes down everyone else the US has control over goes down with them.

China bought 60% of the world chip production and the US and its allies bet that denying those would make China surrender to every US demand. Not only is China not surrendering, it's on its way to domestic production. Notice the things the West says is hard to do, China seems to jump those hurdles. The US banned 5-axis machine tools from China when they were latest and the best and China produced them soon after killing the international market for the US. The US stopped Israel from selling to China AWACS type technology and a couple years later China domestically produced it on its own. They said producing a stealth fighter was difficult. If that were true, China today would still not have a stealth fighter according to predictions of the Pentagon. For decades they said how hard it is to have a successful Mars mission. China on the first try inserted an orbiter, had a lander touch ground, and launched a rover... something no other country has being able to do with Mars.

When the US enacted its strategy to deny chips to China, Chinese domestic production wasn't in their minds because of how difficult they thought it was to accomplish hence why they thought China was going to surrender unconditionally and they were going to win. When people talk about chips and China, they think myopically. They lost 60% of their profits and it's not going to improve because unlike the US, China will not be restricting sales. China will be selling to the world that the competition can't match in costs. They can ban their own companies from buying them... and their products will just cost more making them uncompetitive. Western companies today need the international market to make as much as Chinese companies do just domestically. China is going to be eating away at their 40% and on top of that China will be opening new markets and selling to countries they want to prevent from having this technology. It's just laughable in their propaganda that siding with China, people will have less freedom. Remember Obama's pitch against China's AIIB? He thinks countries want more restrictions and conditions on themselves that want loans for infrastructure that China won't do but the US will. The jackboot on other countries' heads has a label "Made in the USA" on it.

Over on the Beijing Winter Olympics thread, there's discussion on why the South Koreans are so rabidly anti-China. Just like Taiwan, their chip business has no future as well with every power in the world looking to domestic production and they're angry at China and not the US because the Chinese didn't surrender like how they would've if they were in that position. Who are they going to sell their chips to? They can't survive on their domestic market alone. They won't be making world class chips anymore because they have less profits to put into R&D. What countries the US would allow them to sell to... China will be there beating them. It's ironic to side with the US when the US it's all about restrictions. Just look to their own propaganda about countries with less freedom don't survive.
 
Last edited:

tokenanalyst

Brigadier
Registered Member
You are unaware of the dangers. Them dominating production in China is worse than their production elsewhere. Once in China they will now actively suppress any viable competitor. They have to be stopped and let someone else succed.
1-I was just joking with you.
2-How do you know that?
First, if you look at their products they even compete with other companies in annealing and bonding, so far from being a monopoly.
Second, even thought they only sell their more advance products to SMEE, they still sell their products to other companies not only SMEE, so they are helping other high precision equipment makers in China.
Third, like ASML, SMEE need an ecosystem of trusted suppliers and this company for now is the only company in China making this kind of advance equipment. Until now they had delivered what had been entrusted to them.
 

xypher

Senior Member
Registered Member
It is critical to focus on technological node here, SMIC is expanding 14-28nm capacity, not sub 7nm capacity. The latter at the moment cannot be built in China.
They cannot focus on both expanding the 14-28 nm capacity in China and sub 7 nm capacity elsewhere because their budget is limited - they are unable to dish out $40+ billion figures like TSMC since they are still much smaller than them. This is especially true for advanced fabs since the equipment is very expensive, so unless a European partner in JV would be able to weather all the costs, this would inherently hinder SMIC's expansion of Chinese fabs.
I agree that a JV endorsed by European Commision is a must, otherwise the project is doomed to fail.

I don't agree that this would be impossible to achieve. Of course US has strong lobbying in Europe Commission, but as I said there is not a single voice is Europe. The outcome of this attempt, indeed would be very uncertain....it can go both ways, it is impossible to know for sure now.
The thing is that semiconductor supply chains are now deemed a national security level type of thing for many countries, so for any cooperation, there needs to be mutual trust. The latter is on one of the lowest points right now. For example, take CAI which was frozen recently by the EU - it was pretty beneficial for the EU but politics won in the end. I think the same is going to happen with any semiconductor JV, especially if it is going to involve a black-listed entity like SMIC - I think there isn't much incentive in both Europe and China to engage in cooperation on something as important as semiconductors.
I agree that it would be very difficult for Europe to build a 5nm fab all by itself, but this is a very good advantage for SMIC, because it can offer something that Europe alone cannot and this has weight for discussions with EU commission.

Regarding who benefits more, with all the due respect, this is somewhat a bit shortsighted, SMIC should see what advantages they have, and I guess they have a lot, but even more advantages has the vast chinese fabless ecosystem, especially in AI and other advanced technologies because they would have a trusted supplier. In my opinion it is much easier for US to block TSMC or Samsung than blocking an European JV strongly endorsed by European Comission.
There won't be much difference between TSMC, Samsung, or SMIC's JV with a European company for the Chinese fabless companies because the fab would still source equipment from the US and that is covered under the US tech export regulations - meaning that any Chinese company could be arbitrarily black-listed from such a fab just like with TSMC or Samsung.

It is not a jab at the EU btw - the same would happen in China as well if the fab sources the equipment from the US. It is one of the key missions for China is to achieve satisfactory levels of self-sufficiency across the whole fabbing process - from lithography to EDA so that the geopolitical risks from the US sanctions are minimized. Therefore, unless China or\and the EU gain enough self-sufficiency across the whole spectrum any advanced fab would be essentially the same as TSMC\Samsung ones in terms of risks even if the EU and China are on board with it.
Finally a brief comment on times. I think achieving state-of-the-art 5-10 years sooner it makes a lot of difference. What US is trying to do is to slow down China, not to stop it, because they know it's impossible, but for US also slowing down of 5-10 years is a big win, and actually it is. For them it means to have the time to prepare themselves and be ready and reaming ahead of China when China reaches EUV.
The JV fab with a European company is most likely going to take as much time as the current estimates for Chinese EUV, if not longer. It would need to go through the approval process and that is already going to take quite some time with all the political things that are going to inevitably arise and the US pressure. Let's say it passes, picking the location and building the fab is also going to take several years - e.g. TSMC's US fab is expected to finish in 2025 if I remember correctly, and they already started. So for SMIC JV fab, the lead time to establish a 5 nm fab is probably going to be around 2027-2028 if they start right now.

Your proposition is good for discussion and I see your points but I think that at least for now it is not possible to be used in the manner you proposed - as an EU-based fab that would primarily service Chinese fabless companies - because of the reasons I listed above (after the 3rd quote of your post) as the main risk lies not in the TSMC or Samsung as companies but in the US equipment providers.
 
Last edited:

tokenanalyst

Brigadier
Registered Member
This is Open Access. The goal i think is to create a Pulse Laser type light source for EUV using SSMB in a Synchrotron.

Energy recovery linac based fully coherent light source​


Energy recovery linac (ERL) holds great promise for generating high repetition-rate and high brightness electron beams. The application of ERL to drive a free-electron laser is currently limited by its low peak current. In this paper, we consider the combination of ERL with the recently proposed angular-dispersion induced microbunching technique to generate fully coherent radiation pulses with high average brightness and tunable pulse length. Start-to-end simulations have been performed based on a low energy ERL (600 MeV) for generating coherent EUV radiation pulses. The results indicate an average brightness over 1025 phs/s/mm2/mrad2/0.1%BW and average power of about 100 W at 13.5 nm or 20 W with the spectral resolution of about 0.5 meV with the proposed technique. Further extension of the proposed scheme to shorter wavelength based on an ERL complex is also discussed.

An EUV light source based on ERL with the ADM scheme​

The layout of the proposed scheme is shown in Fig.
Please, Log in or Register to view URLs content!
. A DC gun based photoinjector is used to generate high quality electron beam with low emittance, short pulse durations and high repetition rate. Then the electron beam is transported into the main linac and the downstream recirculating loop for the acceleration and radiation. Finally, the electron beam passes through the main linac to recover the energy.
Figure 1
Please, Log in or Register to view URLs content!

The layout of the EUV light source. Round-to-flat (R2F) technique is adopted in the injector.

Please, Log in or Register to view URLs content!
 

gelgoog

Lieutenant General
Registered Member
SMIC isn't investing in 7nm and lower because of the sanctions and restrictions on purchase of Western machine tools. It has nothing to do with lack of capital. For reference an EUV equipped plant costs around $20 billion. That is around twice the cost of their planned fabs in either Shanghai or Beijing.
 

Tam

Brigadier
Registered Member
SMIC isn't investing in 7nm and lower because of the sanctions and restrictions on purchase of Western machine tools. It has nothing to do with lack of capital. For reference an EUV equipped plant costs around $20 billion. That is around twice the cost of their planned fabs in either Shanghai or Beijing.

There is also a lack of demand for 7nm and below, as 14nm and above are used for industrial which is the bulk of use. The sweet spot for industrial applications is around 28nm to 90nm. Reliability is paramount in industrial applications.
 
Status
Not open for further replies.
Top