Chinese semiconductor thread II

Wahid145

Junior Member
Registered Member
We have been seeing a lot of surprises in the Western Countries (including SK) regarding China recently, be it the 6th Gen Fighter, Xiaohongshu, DeepSeek, Unitree (I think these westerners still can't comprehend the Chinese Robotics Development and the consequences of it) or CXMT D1z DDR5.

The only two missing piece of puzzles now are EUVL and DUVi. The amount of research papers @tokenanalyst is showing us here in regards to the EUVL, it feels like the calm before the storm (the storm being innovative products upon products when it comes to EUVL from China in next few years or so). They will literally commoditize EUVL in mid 2030s just like every other thing China grabs a hand on. Can't wait for that tsunami to start!

Sometimes it still feels hard to believe EUVL is the last bastion stronghold the West has when it comes to Choaking China... This country has come a long way in last 8-10 years after being punched in the face by the "most powerful country in the world" with sanctions tariff and all the nasty things with kidnapping the daughter of a CEO etc and etc.
The next thing is going to be Unitree I guess looking at their Robots dancing in Spring Gala. It keeps coming on and on and on relentlessly, The Made In China 2025 plan.

Wonder what Chips these Unitree Robots are using and again, working DUVi EUV Lithography Machine are absolute paramount for this roller-coaster ride to continue in next 5 years solidifying China as Global leader in High Tech!
 

supersnoop

Major
Registered Member
Just a bit of an idea for discussion.
There are so many posts on EUV development every day here, and we see the advances in EV, solar, AI, 5G, robots etc.
Is China really a lot closer to EUV than we realize?
The science is well understood right? Is there really some kind of secret sauce that Chinese companies can't figure out?
Is it possible that Chinese companies are taking so much time because they are looking leap right into sub nm production?

Just from a commercial standpoint, EUV is sub mostly for 7nm production which is just 20% of the market. Samsung is cutting future fabs and we don't have to get into Intel again.

There are not a lot of customers besides Intel, AMD, Apple, Qualcomm, nVidia for these chips. Besides Huawei, I don't think there is a mature enough domestic market of logic designers. Even with great logic designers, they will also need to build the software ecosystem. Perhaps some expectations need adjusting...
 

siegecrossbow

General
Staff member
Super Moderator
Just a bit of an idea for discussion.
There are so many posts on EUV development every day here, and we see the advances in EV, solar, AI, 5G, robots etc.
Is China really a lot closer to EUV than we realize?
The science is well understood right? Is there really some kind of secret sauce that Chinese companies can't figure out?
Is it possible that Chinese companies are taking so much time because they are looking leap right into sub nm production?

Just from a commercial standpoint, EUV is sub mostly for 7nm production which is just 20% of the market. Samsung is cutting future fabs and we don't have to get into Intel again.

There are not a lot of customers besides Intel, AMD, Apple, Qualcomm, nVidia for these chips. Besides Huawei, I don't think there is a mature enough domestic market of logic designers. Even with great logic designers, they will also need to build the software ecosystem. Perhaps some expectations need adjusting...

There is a possibility that they will keep things under wraps until a date of national significance like October 1st.
 

supersnoop

Major
Registered Member
There is a possibility that they will keep things under wraps until a date of national significance like October 1st.

Deepseek's release had me thinking that the "necessary conditions" are not yet present.
If EUV is unveiled tomorrow (or October 1st), who is the customer besides Huawei? Maybe UNISOC?
Moore Threads and Biren don't have designs (or the software environment) that are competitive enough yet.
There is no replacement for Intel yet, EV exports are still in their infancy.
RISC-V is still gestating.
I look at Intel with all their High NA EUV machines and still people are talking about whether they can stay in business. Not only did Samsung cut their productions lines, the recently unveiled Galaxy S25 won't use any Samsung chips at all, so their logic business is probably dead.
 

vincent

Grumpy Old Man
Staff member
Moderator - World Affairs
Just a bit of an idea for discussion.
There are so many posts on EUV development every day here, and we see the advances in EV, solar, AI, 5G, robots etc.
Is China really a lot closer to EUV than we realize?
The science is well understood right? Is there really some kind of secret sauce that Chinese companies can't figure out?
Based on what I read over the years, yes, very much some secret sauce. Say you can produce a high-end gear for some machinery with the exact alloy as your competitor, but your product doesn't last as long as the competitor's because the competitor produce it with a specific process that makes the alloy stronger.
 

sunnymaxi

Major
Registered Member
If EUV is unveiled tomorrow (or October 1st), who is the customer besides Huawei? Maybe UNISOC?
Moore Threads and Biren don't have designs (or the software environment) that are competitive enough yet.
bro what are you talking about ..

SMIC , Huawei and CXMT will be the EUV customers.

you forgot, Loongson , Phytium and obviously Hisilicon in high end fabless category. BYD also entered in high end Automobile chip design sector. Xiaomi also developing its own first high end chip. UNISOC this year record 5G chip sales.. in chip design China is already a world class player.

once domestic advance logic chip production increase you will see intel , Qualcomm loose market share. and i don't need to talk about how Huawei is going to capture Nvidia market share in Ai category.
 

sunnymaxi

Major
Registered Member
Based on what I read over the years, yes, very much some secret sauce. Say you can produce a high-end gear for some machinery with the exact alloy as your competitor, but your product doesn't last as long as the competitor's because the competitor produce it with a specific process that makes the alloy stronger.
that is not true.. Alloy and machine tool component cannot be compare with EUV. you have to achieve certain benchmark in order to manufacture EUV.. for example ''An EUV light source has a wavelength of 13.5 nanometers. This is the standard wavelength used in EUV lithography''. other-wise your machine fell from commercial EUV category.

and you didn't give the answer to @supersnoop 's question.

Lithography is actually a very time consuming project. from R&D to intense testing in lab, then manufacture various components from light source to lens to EUV mirror to Vacuum chamber to wafer handle and so on. you need proper scientific infrastructure and industrial depth. produce first prototype send it to fab for testing and evaluation then receive feedback and do more work on prototype.. now imagine the timeline.. alone testing will take years then scale up production will take time.

we all know the condition of Chinese institutes before sanctions.. they woke up only after 2018-19.

@supersnoop i m sure, you are following the development of EUV on this thread..

there is no secret sauce and everything has been sorted out and successfully developed. teams won provincial award and last year in Summer ''Optics valley conference'' held, in which team disclosed the DUVi/EUV for the first time.

it will still take time for local EUV to enter in production if everything goes well. this is the standard procedure. they have to pass it. no other option.
 
Last edited:

Wahid145

Junior Member
Registered Member
Deepseek's release had me thinking that the "necessary conditions" are not yet present.
If EUV is unveiled tomorrow (or October 1st), who is the customer besides Huawei? Maybe UNISOC?
Moore Threads and Biren don't have designs (or the software environment) that are competitive enough yet.
There is no replacement for Intel yet, EV exports are still in their infancy.
RISC-V is still gestating.
I look at Intel with all their High NA EUV machines and still people are talking about whether they can stay in business. Not only did Samsung cut their productions lines, the recently unveiled Galaxy S25 won't use any Samsung chips at all, so their logic business is probably dead.
This is just rediculous. There's an absurd amount of Chinese Fabless companies dying to fab their Chip in sub 7nm Process node. A lot of companies are/were on the verge of bankruptcy when america stopped China from fabbing Chips in TSMC (Biren, Moores Thread and obviously Huawei and many more). If SMIC can volume produce sub 7nm Chip, it will unleash a lot of more of these "magical" stuff we keep seeing from Chinese companies.

You can think of it this way. One Country has all the resources and don't know how to use them properly (OpenAI, Perplexity, Google, Apple, Quallcomm etc having access to best of the best technology on the planet).

Then there's another country under heavy sanction and limited resources showing wonders to the planet (Huawei 7nm Chip which is on par with 4nm Chip by Quallcomm, DeepSeek showing magic with dumb down GPUs and many more)

Imagine if Huawei can fab their Chip in 3nm process node or DeepSeek and all the rest of Chinese companies have access to the latest NVIDIA chip or Huawei equivalent.

Huawei literally can't make NVIDIA equivalent due to limitation with process node because China don't have EUVL.
 

supersnoop

Major
Registered Member
bro what are you talking about ..

SMIC , Huawei and CXMT will be the EUV customers.

you forgot, Loongson , Phytium and obviously Hisilicon in high end fabless category. BYD also entered in high end Automobile chip design sector. Xiaomi also developing its own first high end chip. UNISOC this year record 5G chip sales.. in chip design China is already a world class players.

once domestic advance logic chip production increase you will see intel , Qualcomm loose market share. and i don't need to talk about how Huawei is going to capture Nvidia market share in Ai category.
You are basically confirming my point though. BYD is "entering". Xiaomi is designing their "first" chip. The market is just starting to develop. Xiaomi hasn't even released an SUV or started international sales of their cars, so the big potential is still to come. You don't actually need EUV until then. If SMIC could produce 3nm chips tomorrow, what would their sales be? Loongson is not pushing the volume of Apple.

I think the followers of the thread feel more urgency in seeing the development than the commercial reality.

that is not true.. Alloy and machine tool component cannot be compare with EUV. you have to achieve certain benchmark in order to manufacture EUV.. for example ''An EUV light source has a wavelength of 13.5 nanometers. This is the standard wavelength used in EUV lithography''. other-wise your machine fell from commercial EUV category.

and you didn't give the answer to @supersnoop 's question.

Lithography is actually a very time consuming project. from R&D to intense testing in lab, then manufacture various components from light source to lens to EUV mirror to Vacuum chamber to wafer handle and so on. you need proper scientific infrastructure and industrial depth. produce first prototype send it to fab for testing and evaluation then receive feedback and do more work on prototype.. now imagine the timeline.. alone testing will take years then scale up production will take time.

we all know the condition of Chinese institutes before sanctions.. they woke up only after 2018-19.

@supersnoop i m sure, you are following the development of EUV on this thread..

there is no secret sauce and everything has been sorted out and successfully developed. teams won provincial award and last year in Summer ''Optics valley conference'' held, in which team disclosed the DUVi/EUV for the first time.

it will still take time for local EUV to enter in production if everything goes well. this is the standard procedure. they have to pass it. no other option.
I have been following, this is why I'm starting to get the sense that EUV is basically "ready very soon". The science behind it is well known, so basically it is a matter of industrial maturity. Can you make these molecularly 'perfect' optics, gratings, chemically pure gases, etc.
 
Top