Text / Li Xin
SMIC has lost a capable person.
On July 4, SMIC issued an announcement stating that Dr. Wu Jingang, the company’s core technical staff, recently applied for resignation due to personal reasons and completed the resignation procedures. After resignation, Dr. Wu Jingang no longer holds any position in the company. At present, the company’s technology research and development work is in normal progress, and the resignation of Dr. Wu Jingang has not had a significant adverse impact on the company’s overall research and development strength.
After the announcement of the news, on July 5, the SMIC Science and Technology Innovation Board saw a drop of up to 1.4%, and finally closed with a slight increase of 0.41%, while the Hong Kong stock market closed down 2.15%, and the market value evaporated nearly 30%. billion Hong Kong dollars.
SMIC’s stock price trend on July 5 (Hong Kong stock), Tuyuan Tiger Securities
This is not the first time that SMIC has experienced the resignation of core talents. Previously, its co-CEO Liang Mengsong also mentioned resignation, but SMIC finally stayed. Although SMIC did not indicate how to keep Liang Mengsong, it can be seen from Liang Mengsong’s salary in SMIC in 2020 that his income has increased by about three times compared with 2019.
It is worth noting that not long before Wu Jingang left, he also received an option incentive worth tens of millions. This sudden departure means that he has given up tens of millions of options.
In addition to the loss of core talents, SMIC is currently facing many challenges in advanced manufacturing processes.
Since its establishment, SMIC’s 28nm, 14nm, 12nm, and N+1 (internal code) technologies have all entered mass production. At the same time, Liang Mengsong also mentioned that the development of 7nm technology has also been completed. In the field of chips, the smaller the process, the lower the power consumption and the higher the performance.
However, SMIC was originally scheduled to risk trial production of 7nm in April, but there has been no news so far, which has also caused concerns about SMIC.
More importantly, the EUV lithography machine originally ordered by SMIC from the lithography machine manufacturer ASML is suspected to be delayed. In March of this year, there was no EUV lithography machine in the re-signed purchase agreement between SMIC and ASML.
EUV lithography machine is the key equipment for advanced manufacturing process. If SMIC has not obtained it, it is difficult for SMIC to make a big breakthrough in advanced manufacturing process, and the gap with TSMC may further widen in the future.
Perhaps realizing the difficulties in advanced manufacturing processes, SMIC has begun to turn to the development of mature processes. But in this regard, SMIC does not have many advantages. After all, there are many manufacturers with mature technology, and it is difficult for SMIC to obtain more orders in the fiercely competitive environment.
In the future, SMIC will not only find ways to break through the advanced manufacturing process and continue to deepen its moat, but also need to expand its revenue channels to prepare for the long-term competition in the future.
1. After giving up tens of millions of incentives and leaving for 20 years, what will SMIC rely on to retain talents in the future?
Among the five core technical personnel of SMIC, Wu Jingang is a truly capable veteran.
According to public information, SMIC was established in 2000, and Wu Jingang joined SMIC in 2001. He previously worked at the Industrial Technology Research Institute of the Ministry of International Trade and Industry of Japan. From 2001 till now, Wu Jingang has served as Assistant Director, Director, Senior Director and Vice President of Technology R&D of SMIC.
Compared with the other five core technicians, Wu Jingang was the first to join SMIC, and he is also the core technician who has worked at SMIC for the longest time. made a lot of contributions.
SMIC’s 2020 Hong Kong stock financial report (the following data are all from Hong Kong stock financial report) shows that Wu Jingang was responsible for participating in the research and development and management of the company’s FinFET advanced process technology during his tenure.
In the financial report released by SMIC’s Hong Kong stock market, the importance of FinFET can be seen, which means that integrated circuit manufacturing needs to be carried out under highly sophisticated equipment. The 35-micron CMOS process is developed to the nano-scale FinFET process.
It can be seen that the FinFET process has become a key technology for integrated circuit manufacturing, and SMIC has entered the second-generation technology development stage in the FinFET process, and has now entered risk mass production.
Risk mass production means that the samples after small batch trial production go through packaging testing, functional verification and other links. If they meet the market requirements, they will enter the stage of risk mass production. The stage of risky mass production mainly includes links such as product yield improvement, production process capability improvement, and production capacity expansion.
It is worth noting that SMIC just announced a new equity incentive plan in May this year. As one of the core technicians, Wu Jingang obtained 160,000 restricted shares, accounting for 0.21% of the total number of restricted shares granted.
The first grant price of the incentive plan is determined based on 36.46% of the company’s average A-share trading price of RMB 54.86 on the trading day prior to the announcement of this plan, which is RMB 20 per share. This means that the market value of the 160,000 restricted shares Wu Jingang obtained is 8.78 million yuan.
However, according to the equity incentive plan, if the incentive object resigns, the restricted stock that the incentive object has been granted but has not vested shall not be vested since the date of resignation, and will be invalidated. That is to say, Wu Jingang gave up options worth tens of millions within less than 2 months after getting 160,000 restricted shares.
Originally, SMIC wanted to “retain people” by options, but I didn’t expect that core technical talents could no longer see tens of millions of options. This may indicate that in the future, SMIC will not be able to retain core technical talents by relying on “money” alone. .
It should be noted that this is not the first core talent resignation storm that SMIC has encountered.
In December 2020, a resignation letter from Liang Mengsong, co-CEO of SMIC, attracted the attention of the whole network. In the letter, Liang Mengsong said, “I feel deeply that I have been disrespected and distrusted. I think you should no longer need me to continue to work hard for the future of the company.”
According to the public information reviewed by Wired Insight, Liang Mengsong became the co-CEO of SMIC in October 2017. In the internal division of labor, Liang Mengsong is responsible for the advanced process, and another co-CEO Zhao Haijun is responsible for the mature process.
Zhao Haijun and Liang Mengsong, picture source SMIC Hong Kong stock financial report
After Liang Mengsong joined SMIC, he quickly showed his ability. In 2018, SMIC’s 14nm process successfully entered the customer introduction stage, and mass production will be achieved by the end of 2019. Since then, SMIC has begun to develop N+1 and N+2 projects, which are equivalent to benchmarking TSMC’s 7nm process.
It can be said that Liang Mengsong is a key player in SMIC’s advanced manufacturing process. Once he resigns, it may be difficult for SMIC to find a person who can carry the banner of breaking through the advanced manufacturing process in the short term.
To this end, SMIC also issued a special announcement at that time, expressing that it had learned of Liang Mengsong’s willingness to resign, and was actively verifying his true willingness to resign with Liang Mengsong.
From SMIC’s 2020 Hong Kong stock financial report, it can be seen that in order to let Liang Mengsong work with peace of mind, it can be said that he has invested “blood capital”.
In 2020, Liang Mengsong’s total compensation was US$4.413 million, including a house of US$3.4 million, which was given to Liang Mengsong by SMIC for home life. That is to say, excluding housing, Liang Mengsong received a total salary of $1.013 million in 2020. Compared with the salary of $341,000 he received in 2019, it has increased by about three times.
However, in Liang Mengsong’s previous resignation letter, he said: “I didn’t come to mainland China to seek high-ranking officials, but simply to contribute to the mainland’s high-end integrated circuits.”
As a big man with more than 35 years of experience in the semiconductor industry, Liang Mengsong does have the capital to say this. During his tenure at Samsung, he helped it to overtake TSMC on the 14nm process, and once grabbed the Apple A9 chip order from TSMC. After joining SMIC, Liang Mengsong also quickly narrowed the gap between SMIC and TSMC in advanced technology.
Although SMIC later left Liang Mengsong, the news of his feud with another co-CEO, Zhao Haijun, was frequently reported, which may also become another major hidden danger for SMIC in the future.
In addition to the rumored discord among executives, the overall turnover rate of SMIC is also high.
According to SMIC’s “2019 Social Responsibility Management Report”, the company’s employee turnover rate was controlled at 17.5% that year, a decrease of 4.5% compared with 2018. Among them, the employee turnover rate in Shanghai, Beijing, Shenzhen, Tianjin and Jiangyin areas They accounted for 45.3%, 32.6%, 12, 3%, 7.8% and 2%, respectively.
SMIC’s employee turnover in 2019, Figure source SMIC’s “2019 Social Responsibility Management Report”
In contrast, the data disclosed by TSMC shows that the employee turnover rate in 2019 was 4.8%. In the five years from 2015 to 2019, the highest employee voluntary turnover rate did not exceed 5%, and the lowest was only 4.1%.
The semiconductor manufacturing industry is a talent-intensive industry, and chip talent is in short supply. The talent challenge SMIC faces has become more and more severe.
2. Waiting for the EUV lithography machine to fail, SMIC’s advanced process research and development is hindered
“At present, technologies such as 28nm, 14nm, 12nm and n+1 have entered mass production, and the development of 7nm technology has also been completed, and it can enter risk mass production immediately in April next year.” At the end of 2020, in Liang Mengsong’s resignation letter , revealed the latest research and development progress of SMIC’s advanced technology.
It is undeniable that since Liang Mengsong joined SMIC, its advanced process technology has indeed embarked on the fast lane.
Before 2017, SMIC’s 28nm low-end Polysion was barely mass-produced, and the mainstream 28nm HKMG yield was not as good as expected.
After Liang Mengsong joined the company, he began to make a series of adjustments, such as strengthening the responsibility system, adjusting and updating the 14nm FinFET plan, etc. He himself also stated in his resignation letter that he has been the co-CEO of SMIC for more than three years and has almost never taken vacations.
Under the leadership of Liang Mengsong, more than 2,000 engineers from SMIC completed five generations of technology development from 28nm to 7nm in three years.
“This is a task that an average company takes more than 10 years to complete.” Liang Mengsong emphasized in his resignation letter.
However, SMIC was originally scheduled to risk trial production of 7nm chips in April this year, but there has been no news until now. This may be related to the delayed arrival of EUV lithography machine, which is considered by the industry as a necessary equipment for chip manufacturing below 7nm.
According to Wired Insight, the current lithography machine is mainly divided into EUV lithography machine and DUV lithography machine, of which DUV lithography machine is also divided into dry type and liquid immersion type. The liquid immersion DUV lithography machine produced by the Dutch lithography machine giant ASML has a wavelength of 193nm, which is equivalent to 134nm.
ASML’s DUV lithography machine, source ASML official website
Although after multiple exposures, the liquid immersion DUV lithography machine can also reach the 7nm process, but one more exposure will greatly increase the manufacturing cost, and it is also difficult to control the yield.
The EUV lithography machine uses a light source with a wavelength of 13.5nm, which is undoubtedly an indispensable device for breaking through the chip process below 10nm. In other words, without ASML’s EUV lithography machine, the 5nm production lines of all chip manufacturing giants cannot be put into production.
It is precisely because of the importance of the EUV lithography machine that SMIC spent US$120 million to order an EUV lithography machine from the Dutch company ASML as early as 2018. The original plan was to deliver it in early 2019. , but has not yet been delivered.
At that time, the Dutch government did issue an export license to ASML, but due to the disguised ban by the United States, the EUV lithography machine was restricted by international agreements and formed export controls, and SMIC was delayed in getting the coveted EUV lithography machine. .
Why a Dutch company would obey the United States actually has its roots.
As early as the 1980s and 1990s, the light source of the lithography machine was stuck at 193nm for 20 years. In order to break through the technology, Intel persuaded the then US President Clinton to organize an EUV LLC alliance, which brought together Motorola, IBM, and the three major national laboratories of the United States.
Due to the technological hegemony between the United States and Japan at that time, the United States did not allow Japanese companies to join, but instead allowed Dutch companies ASML to share research results. To this end, ASML has built a research and development center in the United States, and also ensures that 55% of its components are purchased from the United States and are regularly reviewed.
Up to now, the top 17 suppliers of ASML are mainly concentrated in Europe, the United States, Japan and Taiwan, China, of which 9 are in the United States, 4 in Taiwan, 3 in Japan, and 1 in Germany. In other words, ASML can create EUV lithography machines, and most of its main components are also supplied by the United States.
After SMIC has been waiting for the EUV lithography machine for a long time, it may also realize that there is no hope of delivery.
In March this year, SMIC announced that it had signed a revised and restated bulk purchase agreement with ASML (Shanghai) Electromechanical Equipment Co., Ltd., with a total price of approximately US$1.2 billion, and the agreement was extended to December 2021. March 31.
However, in the announcement issued by ASML, it was shown that the bulk purchase agreement between SMIC and ASML is related to the existing agreement for DUV (deep ultraviolet) lithography technology. That is to say, after waiting for three years, SMIC failed to obtain EUV lithography machine and could only buy DUV lithography machine.
SMIC, which has lost its EUV lithography machine, will only be farther and farther away from TSMC in advanced processes in the future, which cannot be made up by top technical talents alone.
ASML’s EUV lithography machine, source ASML official website
More importantly, according to AI Finance and Economics, there are dozens of machines and hundreds of devices on a chip production line. In SMIC, overseas equipment accounts for about 90%. Among them, American semiconductor equipment accounts for about 60%, the rest are equipment from Japan and South Korea and other countries, and domestic equipment accounts for only about 10%.
This means that SMIC cannot develop advanced manufacturing processes by replacing domestic equipment, because China is relatively backward in the field of chip equipment, and there is currently no equipment that can meet the actual needs of chip manufacturing companies. To this end, SMIC can only achieve breakthroughs in the research and development of advanced manufacturing processes by cooperating with international manufacturers.
But at present, SMIC has neither obtained the EUV lithography machine, the key equipment to break through the 10nm process, nor can it cooperate with domestic manufacturers. If SMIC’s advanced process research and development continues to stagnate, the dilemma it faces can be imagined.
3. The advanced process is not profitable, and the mature process is highly competitive, and SMIC is in an embarrassing dilemma
SMIC’s 7nm trial production has not yet been risked, and TSMC is already considering 3nm commercial production.
According to foreign media reports, TSMC will use the 3nm process technology to manufacture a new factory for chips one year after mass production of the 5nm process technology. The current construction is progressing smoothly and has not been affected.
In the opening speech of the 2021 International Solid State Circuits Conference (ISSCC), TSMC Chairman Liu Deyin also revealed that 3nm is progressing smoothly, even ahead of expectations. The 3nm chips will be trial-produced later this year, with commercial production expected to begin in the second half of 2022.
TSMC’s 18th wafer fab, the main production base for 3nm process, the official website of TSMC
In the face of the rapid progress of TSMC, after the delay in obtaining an EUV lithography machine, SMIC is already advancing the path of developing 7nm without using an EUV lithography machine.
On March 17 this year, SMIC announced that the first generation of SMIC’s FinFET technology has been successfully mass-produced, and the second-generation FinFET technology is in the stage of risky mass production.
Previously, SMIC announced that the first generation of FinFET technology is essentially a 12nm process improved by 14nm process, and the second generation of FinFET technology includes N+1 and N+2 (internal code names), which are benchmarked against TSMC 7nm and 5nm respectively.
However, judging from the officially announced parameters, there is still a certain gap between SMIC’s 7nm and TSMC’s 7nm.
Liang Mengsong once disclosed the situation of N+1 and N+2 generation processes. At that time, he revealed that the N+1 process improved by 20% in performance compared to 14nm, reduced power consumption by 57%, reduced logic area by 63%, and reduced SoC area by 55%. After that, the N+2 process performance and cost are higher.
Although SMIC’s N+1 is very similar to 7nm in power consumption and stability, its performance has not reached the 35% market benchmark improvement that 7nm should have. Therefore, SMIC’s 7nm is not 7nm in the true sense, and it is precisely because the performance improvement is not enough, so the N+1 process is aimed at low-power applications.
It should also be noted that although SMIC has invested heavily in advanced manufacturing processes, the current advanced manufacturing processes have not brought more revenue to SMIC.
According to Wired Insight, starting from the second quarter of 2020, SMIC no longer calculates the revenue share of 14nm chips, which currently represent the company’s most advanced process technology, separately, but discloses it in combination with 28nm chips.
In the first quarter of 2021 financial report, 14/28nm revenue accounted for only 6.9% of SMIC, down 1.9% from the previous quarter. Due to the combined calculation, it is impossible to know the specific amount of 14nm revenue, but from the perspective of the general environment, SMIC’s revenue from 14nm chips is obviously decreasing.
First of all, it needs to be clear that the more advanced the chip technology, the smaller the customer range. In China, the main demander of 14nm and below chips is Huawei HiSilicon. Due to Huawei’s sanctions, SMIC once became a supplier of Huawei’s HiSilicon 14nm chips.
In order to take orders from Huawei, SMIC expanded its 14nm production capacity from 3,000 to 15,000 in early 2020.
But who could have predicted that after SMIC expanded production for half a year, the US sanctions against Huawei would increase. Since September 15, 2020, SMIC and other fabs will not be able to use US equipment to manufacture chips for Huawei. In this regard, SMIC not only lost major customers, but also fell into excess production capacity.
Today, SMIC’s main revenue still relies on mature processes of 40/45nm and 55/65nm, accounting for 16.3% and 32.8% respectively in the first quarter of 2021, which together account for 49.1% of total revenue.
Income analysis by technology node, Tuyuan 2020 SMIC Hong Kong Stocks Financial Report
Unable to make money from advanced processes, SMIC has also turned to expanding mature processes.
On March 17 this year, SMIC announced that it would invest and build a factory in Shenzhen. It is reported that the investment of SMIC’s new Shenzhen project is about 2.35 billion US dollars, or about 15.3 billion yuan. The construction of the plant will focus on the production of 28nm and above processes, and it is expected to start production in 2022, with a planned monthly production capacity of 40,000 pieces.
However, SMIC’s mature technology also shows lack of competitiveness. From 2017 to 2019, its 28nm product revenue accounted for 8.12%, 6.19% and 4.03%, respectively, with an obvious downward trend.
At the same time, in the mature technology, SMIC is also facing more competition. At present, the global 28nm orders are mainly divided by a few companies such as TSMC, UMC, Global Foundry, and SMIC, and SMIC has not received much share.
Today’s SMIC faces losses due to lack of sufficient orders in advanced manufacturing processes, and does not occupy an absolute advantage in mature processes, which can be said to be in a more embarrassing situation.
“If the company’s future investment in technology research and development is insufficient and cannot support the needs of technology upgrades, it may lead to the company’s technology being overtaken or replaced, which will have an adverse impact on the company’s continued competitiveness.” In the financial report risk warning column, SMIC is truthful. show.
Since its establishment in 2000, SMIC has survived 20 years of precariousness and has been in the historical process of industrial development and international relations. In the future, the mission of SMIC will be more important, and the challenges it faces will also be more severe.