Chinese semiconductor thread II

[Hyper]

I know not all, What I'm implying is that they will introduced the first 7nm domestic line.

There needs to be a market for 7nm nodes. Since 3nm is the last finfet node for tsmc they are pushing all customers to shift to 3nm slowly.

 

[Index]

Doing nothing would have been a reasonable alternative had they not boxed themselves into a corner by demonizing China to the public and hyping up an issue that could have been ignored. Government controls media and media influences people and those people put pressure on the government by voting. So, right from the root, they did it to themselves. If they just shut up and didn't say anything to anyone about China, that would have been better. If they were intelligent enough to paint a semi-positive image of China to the American people so they can lay a deeper trap for China increasing our reliance and improving America's image in China by softening their stances on Taiwan and the SCS, that would have lulled China into a very dangerous situation, should they do a sudden 180 and impose all sanctions at once in a conflict we didn't think they would care to escalate. But that's talking about a 4D chess move being played by a kid with Down Syndrome so... we're here instead.

Trying to lay a deeper trap against China would not work because being able to lay a trap in tech for China depends on having a lead over China in technology, while US has parity at best. US is still reliant on SK and even China's semiconductors.

So if they waited until a critical moment and genuinely threatened China's national IC scene (NOT just Huawei), China can just nationalize/ban TSMC from selling to US and instantly bottleneck all of US' needs into Samsung, which they can't sustain. Even worst, given SK's precarious security situation, China can easily manipulate delays/sabotage on Samsung.

In a battle of annihilation, US doesn't have the semiconductor capacity at home to stand directly to China, so they are trying a peeling the cabbage strategy by going after private companies first. But if they are not posing a credible threat to Huawei, there will be no need for Beijing to move out in person and respond.

 

[curiouscat]

Please, Log in or Register to view URLs content!

Please, Log in or Register to view URLs content!

These are some very important insight into Fab expansion.
@curiouscat if you could please link them in pastebin!

SMIC Expansion

Please, Log in or Register to view URLs content!

Hua Hong Expansion

Please, Log in or Register to view URLs content!

 

[tphuang]

https://twitter.com/i/web/status/1829943156353482972

Activist investor forced Texas Instruments to roll back on capex plans for 2026 and onward. This and other pressure on fabs in America will give more room for Chinese fabs expansion. Less capacity from American fabs in analog chip market is positive for all that domestic expansion.

 

[tphuang]

This is old news. SMIC is building 340,000 wafers per month of 28nm capacity with 300mm wafers in four factories.

These factories are named SMIC Jingcheng (100,000 wpm in Beijing), SMIC Lingang (100,000 wpm in Shanghai), SMIC Shenzhen (40,000 wpm in Shenzhen), and SMIC Xiqing (100,000 wpm in Tianjin).

From what I remember SMIC Shenzhen is already in production, SMIC Jingcheng and SMIC Lingang should have the construction of their fab shells complete and should be moving equipment in to start production this year if they have not done it already.

Please, Log in or Register to view URLs content!

This expansion was announced several years ago in like 2020.

But most of this expansion will be done with imported lithography machines. Probably ASML. The only question mark would be with regards to SMIC Jingcheng since the factories financed by Beijing region have higher requirements for Chinese equipment.

FYI, jingcheng fab actually has a phase 2 which will bring eventual capacity to 200k wpm. If we add in smsc sn2, we are looking at about 500k wpm in expansion in total. At some point, they should add 100k wpm per year.

 

[gelgoog]

FYI, jingcheng fab actually has a phase 2 which will bring eventual capacity to 200k wpm. If we add in smsc sn2, we are looking at about 500k wpm in expansion in total. At some point, they should add 100k wpm per year.

Yeah I also read that somewhere. If you look at the satellite pictures there seems to be enough space on that plot of land to double the fab footprint of Jingcheng. The Lingang site also has huge amounts of empty space in it. Enough to double fab space.

The SMIC Xiqing site, if I got it right, seems to be heavily under construction on the satellite pictures I had access to. But the fab shells should be finished like next year I think.

That is a satellite picture of N 38.97877 E 117.19611 in Xiqing, Tianjin. I think that is the site of the future fab but still too early to tell. That picture was taken in Summer 2023.

The Sentinel-2 pictures are blurry as hell and it is hard to tell anything apart.



I cannot tell if it is roofed or if they are still building floors in it. The site also still lacks some of the telltale signs of it being a fab.

This is the concept art of how the SMIC Xiqing fab is supposed to look like. It seems to match with this site.

 

[jli88]

Again bro SMEE is an integrator, if you don't have the necessary component you can't assemble the dammed machine, that was the problem for the last 3 years (2021) until the end user were able to verify and tested and improved the machine , they originally benchmark it with NXT 1980i but the improvement and major breakthrough were able to match the performance of NXT 2000i (the reason for the delay). And also the major component providers like U-precision were able to finished their expansion plan in 2023 and RS LASER were also able to develop a power source at 100kw from the previous prototype of 60 kw, that's a huge improvement and achievement. Now all the planets are align to mass produce the DUVi and it was done so in late 2023.

But didn't someone just mention a couple pages back that SMEE has not had lithography machines in logic fabs, only for packaging?

What is the consensus here about the situation with DUV (SMEE, Huawei etc.) and EUV?

 

[gelgoog]

Depends on what you believe. The SMEE immersion DUV machine is either in testing or in limited production somewhere.
The EUV machine is still at an intermediate technological readiness level where they are trying to get the separate components to have enough performance to make a working demo prototype that is viable for mass production of chips.

 

[gelgoog]

SMIC Lingang construction plan.

Artist concept.

 

[tphuang]

But didn't someone just mention a couple pages back that SMEE has not had lithography machines in logic fabs, only for packaging?

What is the consensus here about the situation with DUV (SMEE, Huawei etc.) and EUV?

I think nobody knows the exact status, except that they are in the fabs themselves and we don't know when or if they have been put into HVM.

The original stated goal of SMIC was to have all domestic 14nm process in SMSC by end of this year. It's unclear to me if they will hit that goal. It's important to consider that SMEE has a high bar to clear here since it needs to get to a certain level of reliability to compete against 1980Di. And Guowang's new factory entered into mass production early this year.

So, it seems like they are producing and selling front-end scanners. For what process and at what fab, that's hard to say. And even if they are not being used right now in 28nm node, I'm not sure that's a problem.

People just need to relax about this. EUV is the key here. SMEE is whatever.