Chinese semiconductor thread II

tphuang

Lieutenant General
Staff member
Super Moderator
VIP Professional
Registered Member
Giantpanda and proelite are correct -- "high end" is rather vague. If the original rumour was also vague then that's fine, but that should be the natural first question or at least something to acknowledge.

Is it meeting the threshold of sub 14nm, or sub 7nm or even sub 5nm etc.



Of course, further domestic self sufficiency for the semiconductor production line is something that has been predicted for a long time, the real question is if there is some sort of milestone that's been claimed to have been attained, what actual milestone is it?

The question after that, is what would the speed of implementation and upscaling be
Keep in mind that SMIC wanted to have fully domestic 14nm line by end of this year. And I think that will allow it to move all old 14nm lines to 7nm . Which would give it about 20k wpm of 7nm. For the time being, getting the full 35k wpm at sn1 is the priority, but 50k wpm of finger by end of 2025 isn’t out of question.

we can’t get to real high end stuff until euv comes along. And that’s a couple of years away.
 

CMP

Senior Member
Registered Member
Giantpanda and proelite are correct -- "high end" is rather vague. If the original rumour was also vague then that's fine, but that should be the natural first question or at least something to acknowledge.

Is it meeting the threshold of sub 14nm, or sub 7nm or even sub 5nm etc.



Of course, further domestic self sufficiency for the semiconductor production line is something that has been predicted for a long time, the real question is if there is some sort of milestone that's been claimed to have been attained, what actual milestone is it?

The question after that, is what would the speed of implementation and upscaling be
From a layman rumormonger/journalist's perspective, high end would mean something that can be used to produce flagship cellphone chips and/or AI chips.
 
  • Like
Reactions: jwt

Blitzo

Lieutenant General
Staff member
Super Moderator
Registered Member
Keep in mind that SMIC wanted to have fully domestic 14nm line by end of this year. And I think that will allow it to move all old 14nm lines to 7nm . Which would give it about 20k wpm of 7nm. For the time being, getting the full 35k wpm at sn1 is the priority, but 50k wpm of finger by end of 2025 isn’t out of question.

we can’t get to real high end stuff until euv comes along. And that’s a couple of years away.

Yup, I'm aware of the projected timelines and developments that we are expecting this year etc.

I'm more saying that the rumour itself is sufficiently vague that it can mean anything, such as what CMP wrote above like having full domestic capability to produce flagship phone SoCs and leading edge AI GPUs (even if perhaps they are not yet in trial production), as small as 3nm for example -- for example if EUV was considered having a baseline of demonstrability.

Alternatively, we know that it could mean full domestic 14nm line readiness which would still be a milestone and practical benefit for the reasons you describe, or anything in between.


It just feels like the rumour itself is deliberately phrased vaguely, which is understandable but may not tell us anything new.
 

tphuang

Lieutenant General
Staff member
Super Moderator
VIP Professional
Registered Member
Yup, I'm aware of the projected timelines and developments that we are expecting this year etc.

I'm more saying that the rumour itself is sufficiently vague that it can mean anything, such as what CMP wrote above like having full domestic capability to produce flagship phone SoCs and leading edge AI GPUs (even if perhaps they are not yet in trial production), as small as 3nm for example -- for example if EUV was considered having a baseline of demonstrability.

Alternatively, we know that it could mean full domestic 14nm line readiness which would still be a milestone and practical benefit for the reasons you describe, or anything in between.


It just feels like the rumour itself is deliberately phrased vaguely, which is understandable but may not tell us anything new.
Even if we get to 7nm over the next year, that would be way beyond expectations as far as I'm concerned.
Also, there is the question if this is for risk production or for HVM.
If they were out of access to ASML DUVi today, I'm sure the SMEE can be thrown into action. But whether or not it makes sense to have an all domestic 7nm HVM line is hard to say. From that sense, you probably will see 14nm or 7nm domestic line before 28nm, since the latter still has access to more reliable tools. And the former is so limited, that even lower yield is better than nothing at all
 

tokenanalyst

Brigadier
Registered Member

Huawei's Shanghai Qingpu project has been completed with an investment of over 10 billion yuan​


According to the official website of the Shanghai Municipal Government, on July 9, Huawei's Shanghai Qingpu project has been completed and officially named "Huawei Lianqiu Lake R&D Center". At present, the internal roads, small trains, and elevated overpasses of the R&D center are all connected; the Moon Bridge, Rainbow Bridge, Arch Bridge and other bridge projects across Lianqiu Lake are in the final stage; the greening project in the corners of the park is in the final stage, and Lianqiu Lake has begun to store water.

According to previous news, the R&D center is located in Xicen Community, Jinze Town, Qingpu District, covering an area of 2,400 acres, with a total construction area of 2.06 million square meters and a total investment of over 10 billion yuan. After completion, it will become Huawei's largest R&D center in the world, bringing together diversified functions such as corporate office, R&D pilot, technology incubation, production services and supporting residence, aiming to create a complex industrial community. As Huawei's key R&D base, Huawei Qingpu R&D Center will undertake R&D tasks in the fields of terminal chips, wireless networks and the Internet of Things, and will have 35,000 scientific and technological R&D personnel stationed in the future. To this end, a series of supporting facilities such as single apartments and family housing are specially prepared for talents.

In the future, this will become a core node in Huawei's global R&D network, attracting top talents, nurturing cutting-edge technologies, and helping Huawei make greater breakthroughs in key technology fields such as 5G, cloud computing, and artificial intelligence, and contribute to the progress of the global technology industry.

Please, Log in or Register to view URLs content!
 

tokenanalyst

Brigadier
Registered Member

Tongfu Microelectronics: Net profit in the first half of the year is expected to increase by 253.44%-299.79% year-on-year​

Tongfu Microelectronics issued an announcement stating that the company expects to achieve a net profit attributable to shareholders of listed companies of 288 million yuan to 375 million yuan in the first half of 2024, a year-on-year increase of 253.44% to 299.79%; the net profit after deducting non-recurring gains and losses is 265 million yuan to 355 million yuan, a year-on-year increase of 201.37% to 235.80%.

Regarding the reasons for the performance growth, Tongfu Microelectronics stated that in the first half of 2024, the semiconductor industry showed a recovery trend, market demand rebounded, and technologies and applications such as artificial intelligence promoted the development of the industry. Against the above industry background, the company was proactive, capacity utilization increased, and operating income increased significantly, especially the operating income of mid-to-high-end products increased significantly.

According to preliminary statistics from the company's financial department, Tongfu Microelectronics' operating income in the second quarter of 2024 increased significantly compared with the second quarter of 2023 and compared with the first quarter of 2024. At the same time, thanks to the strengthening of management and cost control, the company's overall efficiency has been significantly improved.

Please, Log in or Register to view URLs content!
 

gelgoog

Brigadier
Registered Member
The translated text isn't that informative. If it wasn't for me knowing that Tongfu is one of the major chip packaging companies I wouldn't even have figured out what the relevance of the article was.
 

tokenanalyst

Brigadier
Registered Member

130um! The world's thinnest silicon carbide wafer is launched​


Jiangsu General Semiconductor Co., Ltd. (formerly: Henan General Intelligent Equipment Co., Ltd., hereinafter referred to as "General Semiconductor") disclosed on July 12 that the company successfully stripped out 130um thick ultra-thin SiC wafers on July 10, 2024 using its self-developed equipment (silicon carbide ingot laser stripping equipment).

1720813142698.png
8-inch SiC ingot laser fully automatic stripping equipment (source: General Semiconductor)
According to information, General Semiconductor was established in 2019 and is committed to the research, development and manufacturing of high-end semiconductor industry equipment and materials.

In terms of financing, Universal Semiconductor completed its angel round and A round of financing in August 2021 and August 2023, respectively. Investors included Tianyan Fund, Lhasa Chuyuan, Hunpu Investment, Northeast Securities, Dingxin Capital and other institutions.

In terms of products, General Semiconductor developed the first domestic semiconductor laser invisible cutting machine in 2020; successfully launched the first domestic 18-nanometer and below SDBG laser invisible cutting equipment (for 3D Memory) in 2022; successfully developed the first domestic 8-inch fully automatic SiC ingot laser stripping production line in 2023; and successfully developed the SDTT laser invisible cutting equipment (for 3D HBM) in 2024.​

Please, Log in or Register to view URLs content!
 

ansy1968

Brigadier
Registered Member
From Motif (China Military Forum) @horse bro you should ask him to join this forum.

It immensely help to propel Chinese semiconductor industry forward. Last year when Mate 60 Pro was launched, the Kirin 9000S chip it used was fabricated with a de-Americanized production line.

Around the time Pura 70 Pro was launched, ie. about two months ago, the domestic production line has completed and came online. By complete, I believe it mean certified in real world production. Which suggest that the Kirin 9010S chip used in Pura 70 Pro could be fabricated with full domestic process, including the domestic lithographic machine.

About two months ago, there was an analysis of the Kirin 9010 chip. It was found that 9000S is clearly manufactured using ASML machine as the Athena AH74 scribeline mark is clearly present. But on the 9010 die, there is none of the ASML scribeline mark, from ancient TTL to Athena and SMASH.


Not only ASML machine, the LSA, FIA scribeline mark of the Nikon, TTL, OAL of the Canon are also not found on the 9010 die.


The conclusion at that time is, there's no evidence that 9010 is fabricated by ASML lithographic machine. What lithographic machine is used is unknown. Now with the revelation of the domestic process being completed around that time, it's very likely to be a domestic lithographic machine.


The 9000S and 9010S are manufactured with the same 7 nm process, they shared a lot of commonality, all components are exactly the same except the CPU cluster. Why would 9000S has clear ASML scribeline mark while 9010 has none? The simplest explanation is it wasn't fabricated with ASML, Nikon or Canon lthographic machine. Then the question is what machine is used?



The US not only banned ASML from selling EUV lithographic machine. Later on they also banned DUV machine. Then later on they pressured ASML to stop maintenance support of existing DUV machine bought before the ban. It’s easy to understand the need of not only de-Americanized process, but a full domestic process. Now they can talk more openly about what chip are used in Huawei phones.
 
Top