Chinese semiconductor thread II

[european_guy]

You can read more of about this in my recent article -

Please, Log in or Register to view URLs content!

ASML Q2 2024 Earnings Preview And 2024-2025 Tailwinds​

Thanks for the report.

I took the liberty to post here below one of your tables:




I would like to highlight the so called "Others" row, that I assume mainly refers to Chinese SME manufacturers.

In particular here below I show the domestic ratio by quarter, i.e. how much of the total equipment revenues in China can be extrapolated to be from Chinese manufacturers.

​	Big foreign SME​	Others (mostly Chinese)​	Domestic ratio %​
Q1 2023​	4357​	1603​	26,9%​
Q2 2023​	5117​	2433​	32,2%​
Q3 2023​	8625​	3757​	30,3%​
Q4 2023​	8656​	3473​	28,6%​
Q1 2024​	8484​	4036​	32,2%​

Of course Others is not exactly equal to domestic SME, but we can assume that it is for the largest part.

We can see that ratio is growing YoY, but not so quickly....it is a hard battle!

 

[tphuang]

Thanks for the report.

I took the liberty to post here below one of your tables:


View attachment 132433

I would like to highlight the so called "Others" row, that I assume mainly refers to Chinese SME manufacturers.

In particular here below I show the domestic ratio by quarter, i.e. how much of the total equipment revenues in China can be extrapolated to be from Chinese manufacturers.

​	Big foreign SME​	Others (mostly Chinese)​	Domestic ratio %​
Q1 2023​	4357​	1603​	26,9%​
Q2 2023​	5117​	2433​	32,2%​
Q3 2023​	8625​	3757​	30,3%​
Q4 2023​	8656​	3473​	28,6%​
Q1 2024​	8484​	4036​	32,2%​

Of course Others is not exactly equal to domestic SME, but we can assume that it is for the largest part.

We can see that ratio is growing YoY, but not so quickly....it is a hard battle!

It’s quite apparent that several Chinese fabs, especially cxmt really ramped up buying of American equipment and asml machines in the past years to stock up before major expansion as well as fear of further sanctions.

it’s unclear to me if this will keep up past this stretch. Or if this is just a temporary trend until domestic equipment makers are caught up in reliability to foreign equipments and scaled up production. After all, domestic production can only climb so quickly

 

[gelgoog]

At least during this year, maybe next year as well, the large tools purchases should continue. SMIC has three 28nm megafabs going up. CXMT and YMTC are doubling their floorspace. Hua Hong is still expanding its Wuxi fab. Nexchip is supposed to ramp up 28nm production.

Beyond 2024, maybe 2025, I doubt there will be that many purchases. CXMT and YMTC still have enough land area to do another expansion. But SMIC will be loaded with debt. They are already spending more in capex than in actual revenue. Hua Hong's expansion into 55nm is being a bit of a bust because of Nexchip bringing the prices down. At the same time when Hua Hong and Nexchip go into 28nm they will be competing with SMIC's new gargantuan capacity. So they are stuck.

There is a possibility the smaller vendors like CRMicro will pick up some of the slack and expand their own operations. But it is quite likely the market will be operating below capacity and there will be no room for them to move in. Unless they get some captive clients or the Chinese government mandates higher chip content with some sort of state support for Chinese chips I do not see this situation changing.

To further increase capacity in the mid term will, I think, require the entrance of new players in the memory market, or for Chinese immersion lithography tools to become available thus allowing the expansion of FinFET capacity.

 

[AndrewS]

At least during this year, maybe next year as well, the large tools purchases should continue. SMIC has three 28nm megafabs going up. CXMT and YMTC are doubling their floorspace. Hua Hong is still expanding its Wuxi fab. Nexchip is supposed to ramp up 28nm production.

Beyond 2024, maybe 2025, I doubt there will be that many purchases. CXMT and YMTC still have enough land area to do another expansion. But SMIC will be loaded with debt. They are already spending more in capex than in actual revenue. Hua Hong's expansion into 55nm is being a bit of a bust because of Nexchip bringing the prices down. At the same time when Hua Hong and Nexchip go into 28nm they will be competing with SMIC's new gargantuan capacity. So they are stuck.

There is a possibility the smaller vendors like CRMicro will pick up some of the slack and expand their own operations. But it is quite likely the market will be operating below capacity and there will be no room for them to move in. Unless they get some captive clients or the Chinese government mandates higher chip content with some sort of state support for Chinese chips I do not see this situation changing.

To further increase capacity in the mid term will, I think, require the entrance of new players in the memory market, or for Chinese immersion lithography tools to become available thus allowing the expansion of FinFET capacity.

I think the most likely scenario is that Chinese companies (with government subsidies) will continue to design out foreign semiconductors with domestic Chinese semiconductors.

Remember that in the mature (DUV) semiconductor market, China currently accounts for 31% of global production and this is expected to increase to 37% by 2027. Yet today, China today accounts for 50-60% of global semiconductor demand, with about half consumed domestically and half exported as raw chips or embedded in products.

So even in 2027, there is still huge scope for Chinese fabs to build out more capacity and replace imported chips, given suitable Government subsidies.

 

[gelgoog]

I wonder what is the situation with chip import substitution sector by sector. Let's say appliances, consumer electronics, cars, smartphones, etc.

 

[siegecrossbow]

Please, Log in or Register to view URLs content!

1TB SSD from Huawei.

 

[tokenanalyst]

The second phase of the National Big Fund invested in Chongqing Xinlian Microelectronics, a specialty process wafer fab​

Tianyancha shows that on July 13, Chongqing Xinlian Microelectronics Co., Ltd. underwent a number of industrial and commercial changes, including the addition of new shareholders Chongqing Manufacturing Transformation and Upgrading Private Equity Investment Fund Partnership (Limited Partnership) and National Integrated Circuit Industry Investment Fund II Co., Ltd.

After the new shareholders have invested, the shareholders of Xinlian Microelectronics include Chongqing High-tech Zone Intelligent Manufacturing Industry Research Institute, Chongqing Xiyong Microelectronics Industrial Park Development Co., Ltd., National Integrated Circuit Industry Investment Fund II Co., Ltd., Qingling Motors (Group) Co., Ltd., and Chongqing Manufacturing Transformation and Upgrading Private Equity Investment Fund Partnership (Limited Partnership).

According to official news from Xinlian Microelectronics, Chongqing Xinlian Microelectronics Co., Ltd. is positioned as a leading specialty process wafer factory in the western region, aiming to build an industry-leading automotive-grade chip manufacturing company. It is a 12-inch high-end specialty integrated circuit process line project with national capital. The planned production capacity of the first phase of the project is 20,000 pieces. It is committed to becoming an important strategic backup for the country's integrated circuits in the west and ensuring the security of the integrated circuit industry chain. The company is located in Chongqing Xiyong Microelectronics Industrial Park. It is a key project built by the Chongqing Municipal Government and the High-tech Zone Government in conjunction with the development of Chongqing's "33618" modern manufacturing cluster system. It is a key project for the transformation and upgrading of Chongqing's high-end manufacturing and integrated circuit industries.

Please, Log in or Register to view URLs content!

 

[tokenanalyst]

Wuxi Disi high-end mask production line officially launched, the first set of 90nm products delivered successfully​

Recently, Wuxi DIS Microelectronics Co., Ltd. (hereinafter referred to as "Wuxi DIS") completed the installation and commissioning of key equipment for its high-end mask project, successfully completed the production line, and completed the production and delivery of the first set of 90nm high-end mask products, marking a new leap in Wuxi DIS's technological capabilities and a solid step towards becoming an open mask company with leading technology and production capacity in mainland China.

Since 2021, Wuxi DIS has seized the opportunities of industry development and achieved rapid growth in performance. At the same time, it has successfully implemented two rounds of financing and invested about 2 billion yuan in Wuxi High-tech Zone to build a high-end mask project. Since the groundbreaking ceremony at the end of November 2022, after 19 months, major milestones such as plant capping, equipment moving in, process debugging, and production line connection have been completed. In the second half of 2024, Wuxi DIS will focus on mass production of 90nm process, complete 40nm mass production in 2025, and achieve 28nm upgrade in 2026, continuously enhancing core competitiveness. After the project is fully operational, the production capacity of 90nm-28nm high-end masks will be increased by 2,000 pieces/month, and the total capacity will reach 5,000 pieces/month.

At the ceremony of "Disi High-end Mask Production and Successful Delivery of the First Set of 90nm Products", Lv Jian, General Manager of Wuxi Disi, said, "Mask manufacturing is a key link in the semiconductor industry chain and a link between chip design and manufacturing. The localization rate of mid-to-high-end semiconductor masks is low and there is a lot of room for growth. Promoting the construction of high-end mask projects is conducive to giving full play to the important roles of 'making up for shortcomings', 'forging strengths' and 'filling in gaps', igniting the engine of high-quality development with new quality productivity, and further consolidating Disi's leading position in the integrated circuit mask segment track."

Please, Log in or Register to view URLs content!

 

[tphuang]

I wonder what is the situation with chip import substitution sector by sector. Let's say appliances, consumer electronics, cars, smartphones, etc.

auto is probably the slowest. smartphones, they are going to get to close to all domestic phone with Mate 70. But for other OEMs, they are going to trail that big time.

The thing is that domestic market chip demand is so high that even the growing domestic production can't catch up

 

[tokenanalyst]

The Institute of Microelectronics has made new progress in high-performance phase-locked loop chips​

Recently, the 2024 IEEE Symposium on VLSI Technology & Circuits was held in the United States. The team of Researchers Li Bo and Yang Zunsong from the Key Laboratory of Radiation-Resistant Device Technology of the Institute of Microelectronics presented the latest research progress of high-performance phase-locked loop chips at the meeting.

The explosive growth of network data interaction has prompted the continuous advancement of communication technology. New generation communication systems such as 5.5G, 6G, and 224Gb/s high-speed interface circuits require the RMS jitter of the phase-locked loop frequency synthesizer to be less than 50fs, which is extremely challenging in silicon-based CMOS integrated circuit design.

The phase-locked loop of the sub-sampling phase-locked loop does not require the use of multi-mode dividers and retimers, and has the potential to further improve jitter performance at low power consumption. For extremely low RMS jitter below 50fs, the buffer in the reference path usually greatly limits the in-band phase noise of the sub-sampling phase-locked loop, requiring a large amount of power consumption and area to reduce it to a lower level. Double-edge sampling phase-locked loop technology can reduce the phase noise introduced by the reference buffer by 3dB, but it will deteriorate the frequency resolution. In response to the above problems, the research team of the Key Laboratory of Radiation-Resistant Device Technology proposed a configurable double-edge sub-sampling phase-locked loop structure. By switching between odd and even sampling modes, it can maintain the frequency resolution while achieving low power consumption and low jitter simultaneously. The proposed phase-locked loop structure has been verified on a silicon-based 65nm CMOS process, achieving the best reference frequency normalized figure of merit for jitter-power consumption in the world.

    The above research results were published in the top international conference in the field of integrated circuit design, 2024 VLSI Technology and Circuits, with the title “A 6.5-to-6.9-GHz SSPLL with Configurable Differential Dual-Edge SSPD Achieving 44-fs RMS Jitter, −260.7-dB FOM Jitter , and −76.5-dBc Reference Spur” . Chen Tianle and Ren Hongyu, doctoral students of the Institute of Microelectronics, are the co-first authors of the article, and Yang Zunsong and Li Bo, researchers of the Institute of Microelectronics, are the corresponding authors of the article.

Please, Log in or Register to view URLs content!