Image source @Visual China
Text丨Dianjin Financial Editorial Department, Author丨Liu Yajie, Editor丨Yan Rui
The slogan of independent research and development has been shouted for so many years, and when Huawei needs help the most, she still retreated; she was short of money, and she remembered the goodness of the capital market. SMIC (00981.HK), which has been listed on the Hong Kong stock market, came to the A-share Science and Technology Innovation Board to sweep away 20 billion yuan of cash.
is like this. On June 19, SMIC passed the and it only took 19 days, which was awesome. The face of the supplier
needs to be seen, the expectations of the Chinese people must be supported, the development of the industry must be carried, and the money of investors must be taken... No matter how you look at it, SMIC is like a young wife who is angry at home. She wants to do anything and can't afford to offend anyone.
Therefore, there will always be extreme spectators, who are pretentious over and over again - "Whatever you can't do, you can do it, you can do it with money." Faced with such evaluation, SMIC is still calm and rational. When quarreling with noise, the same noise that you make yourself.
However, silence does not mean emptying your mind, and listing should not just take money and leave. The fiasco of SMIC this time of capital schema, how to leave your own traces, and what kind of SMIC story needs to be told, is the key.
SMIC is known as the strongest semiconductor foundry manufacturer in China, but whether it grows depends on whether ASML (Asmai, Netherlands) is in the mood to export lithography machine .
"cannot wafer wafers without ASML" and "cannot produce Huawei 14nm Kirin SoCh in the short term". These problems are unsolvable for the time being and are acceptable. The key is that the 20 billion yuan will be spent. After several generations of SMIC , if the gap with TSMC can be shortened, everything is worth it.
"The mainland only needs to engage in chip design, but does not need to enter the field of chip manufacturing." In an interview with a certain TV at the beginning of the year, the outside world interpreted the attitude of Zhang Zhongmou, the founder of TSMC , as an advice, and now I feel more about warnings. breaks the arrogance of TSMC and reflects the confidence of the development of the semiconductor industry of a major country. This is the responsibility of SMIC .
People often overestimate changes in the short term and underestimate changes in the future. I hope that SMIC 's hiding strength and maintaining a low profile is not only for small profits in front of you, but for a bigger future.
01, anti- TSMC front
In the traditional impression, Zhang Zhongmou, who is known as the "Godfather of Taiwan's semiconductors", has a childish face, kind and kind, but in fact there are countless ruthless and tricky inside, and all the points are SMIC .
Before creating SMIC , Zhang Rujing founded Taiwan semiconductor company World Group Circuit Company (WSMC), which was founded in three years, will achieve profitability. However, such a high-quality company has collectively "rebounded" by the management. Zhang Rujing took advantage of the collective vote on the TSMC . There are reports that it was Zhang Zhongmou's "big deal".
was so angry that Zhang Rujing came to Shanghai in 2000 to start a second business. Faced with the increasing SMIC , Zhang Zhongmou's dormitory was uneasy, so he filed four lawsuits between 2003 and 2006, and finally drove Zhang Rujing out of SMIC in 2009. In desperation, Jiang Shangzhou, the former chairman of SMIC, invited professional manager Wang Ningguo to join the temporary helm, pushing Zhang Rujing away from the trend.
Wang Ningguo was lucky. After all, he was not facing a mess full of holes, but a large factory with 12-inch and 8-inch wafer fabs in different sizes in Shanghai, Beijing, Tianjin and other places.
Therefore, during the two years in office, Wang Ningguo was mainly busy sorting out the management structure and stabilizing traditional users. gradually cultivated a company with revenue of US$1 billion in 2009 into a semiconductor company with revenue of more than US$700 million in the first half of 2011 and turned losses into a semiconductor company with revenue of more than US$6.5 million (net profit).
However, at this time, the means of TSMC to "disassemble" SMIC began to work.Due to the dispersed equity, the major shareholder Datang Telecom began to compete with overseas shareholders for power, gradually causing Wang Ningguo to lose his foothold and chose to resign in July 2011. Before him, more than 90 Taiwanese management teams had left. After
, Qiu Ciyun officially took over. He has been immersed in the semiconductor industry for 27 years. He has worked at the Institute of Solid Technology in Munich, Germany and ATT Bell Laboratories in the United States. He is also an old world of TSMC . He relies on his life to promote the development of SMIC .
According to Qiu Ciyun's plan, SMIC cannot be a semiconductor workshop, but must become an "international and independent company." However, judging from the results, the independent effect may be OK, but the trend of internationalization is becoming less and less obvious.
After Qiu Ciyun entered the cabinet, he continued to compress the company's radical external investment and his style was no longer sharp. : Public data shows that in the five years since 2011, SMIC's capital expenditure has decreased year by year, compared with TSMC's capital expenditure is less than 20%.
At the same time, SMIC launched inclusionary development, and its business focus was adjusted from North America to China. Financial report data shows that in 2011, SMIC revenue accounted for 55% and 32.7% of the revenue in North America and China, respectively; by 2016, the proportion of the two places reversed to 29.5% and 49.7% - there was no complete shocks during the Zhang Rujing period.
cools down the company, SMIC reduces internal disputes, and at the same time eliminates the vigilance of TSMC in exchange for stages of peaceful coexistence. However, convergence does not mean disarming. After many years of violation, SMIC still showed its dagger. In May 2017, Liang Mengsong, who worked for IBM, TSMC , and Samsung Electronics , joined SMIC .
In a sense, Liang Mengsong's life experience in was laid out in controversy: Shortly after joining TSMC , he successfully led the development of the 130nm process process, defeating his old boss IBM; after joining Samsung Electronics, he successfully led the development of the 14nm process process, pressing TSMC under his body; after joining SMIC , he achieved the upgrade of 28nm to 14nm technology within 300 days, with a yield rate of 95%.
At this time, the acceleration of SMIC did not attract much attention from TSMC . After all, two years ago, TSMC had been upgraded, which led to Zhang Zhongmou's ridicule. But so what? Bear with it, after all, it is a fact that no one can change it.
02, 20 billion yuan listen to it?
In order to fill the gap as soon as possible, a few days ago, the application for listing of SMIC on the Science and Technology Innovation Board was accepted, and its quota of 20 billion yuan raised also set a record for financing on the Science and Technology Innovation Board. It would be astronomical to give this money to other companies, but for SMIC , if you think of catching up with the goal of TSMC , the money is not much.
In 2003, Zhang Rujing increased the revenue scale of SMIC to US$365 million, and the annual revenue growth rate of 6.3 times has also become a good story. However, after TSMC "torn" a bit, SMIC began to seek stability, and there was no previous speed again.
From 2019, SMIC's revenue was 22.018 billion yuan and a net profit of 1.269 billion yuan. The external pressure continued to increase, and it still recorded nearly 8 times growth, which is considered a remarkable achievement. However, during the same period, TSMC has grown from a company with annual revenue of US$5.807 billion to a giant beast with annual revenue of US$35.774 billion (252.897 billion yuan), and a net profit of US$11.836 billion (about 83.672 billion yuan). Once the size of
is too large, the data will be difficult to measure the real gap.
SMIC prospectus shows that in the past three years, R&D investment has always remained at 20%, which is extremely valuable when other mainstream foundry manufacturers are below 10%. In contrast, TSMC's R&D investment in TSMC has always been less than 10%, but it still costs about 20 billion yuan per year.
SMIC is working hard to raise a record amount of funds, and prepare all the funds spent seven times a penny, just one year of TSMC . "The 20 billion yuan financing of the Science and Technology Innovation Board is too small, and 200 billion yuan is not much, and it also needs to share costs on a large scale."Guosin Securities analyst He Lizhong said.
At present, TSMC cannot resist external pressure and finally decided to give up the OEM solution for Huawei. This is a rare opportunity for SMIC . You should know that SMIC 's main business is concentrated in 40/45nm, 55/65nm, and 0.15/0.18μm, and 14nm/28nm accounts for only 4.32% in total. Cooperating with Huawei is likely to change the existing revenue structure and get rid of the dependence on mid- and low-end businesses.
's idea is always good, but that's just thinking about it. Besides, in-depth discussion of losses, TSMC is far more "hard" than SMIC : TSMC 's lifeline is in the high-end process , 7nm process wafer accounts for 35% of the total shipment volume, and is widely used in Huawei's main models.
's such a large order can stop at any time, which may even further affect 5nm wafer production. Who knows the pain? When TSMC has to let go of cooked ducks, and even SMIC at the door is not qualified to talk about competition and losses, everyone is a leaf floating in the wind and rain.
In this context, SMIC should consider how to protect itself, not whether to save Huawei.
In the short term, SMIC still has room for breathing. By developing the N+1 process, SMIC will improve the performance of 14nm by 20%, reduce the power consumption by 57%, and reduce the SoC size by 55%. The performance of the N+2 process released at the same time remains unchanged, and the stability is further improved. No A SML provides EUV (extreme ultraviolet lithography) equipment, and SMIC will not be discontinued.
, but after 2020, what should I do in the future? Taking the high-end development route, SMIC will sooner or later face TSMC , but lack of lithography machines, 7nm and higher process solutions cannot be started, which is a very difficult and must-solve problem.
" SMIC can reach 7nm without EUV, and then 5nm and 3nm must have EUV. "Liang Mengsong said.
According to the roadmap of TSMC , it will officially mass-produce 5nm in 2020, and plans to 4nm process in 2023, but the technology of SMIC can only be infinitely close to 7nm, and it will never take the step of overtaking. The more difficult the situation is, the more we cannot disarm. Sleeping on firewood and tasting gall is the only choice. When
participated in the 2019 SMIC technical seminar, SMIC joint CEO Zhao Haijun showed its positioning, "Be a fast follower and be the first in the second echelon", this is a very suitable positioning for SMIC . Surviving in the cracks is so bitter and helpless.
03, next Huawei?
endure humiliation and eventually rise against the trend. This is indeed difficult, but such cases in China are not uncommon, Huawei is a good case. So will SMIC become the next Huawei? It is difficult, but it may not be possible.
unless it is day The descendant weapon, otherwise as a chaser, the key to overtaking is to form a speed difference: on the one hand, you run fast enough, and on the other hand, your opponent does not run fast enough, and with the accumulation of time, there will always be a chance to tie or even overtake.
From experience, it is too difficult to descend the development laws of the existing semiconductor industry and turn to find a new lord in troubled times. For the past 20 years, people have been exploring and have not found the answer.
In 2004, the University of Manchester in the United Kingdom discovered graphene, hoping to replace silica, but there is still no progress so far; at the end of 2018, the country re-established The major project "Super Resolved Lithography Equipment Development" has passed the acceptance, increasing the domestic lithography resolution to 22nm, but it has not been commercialized on a large scale.
for so many years, the "magic weapon" that can reverse the situation has never appeared. Even if there is such a turning point, it will have equal opportunities for old forces such as TSMC , Samsung Electronics, Intel , and Grossfield. It is not enough to support SMIC to stand out.
From the perspective of SMIC , it is of little significance to continue to motivate growth. From 2003 to 2019, SMIC's revenue increased by 7 times, while TSMC's only completed 6 times; R&D investment remained at a high level all year round, and 8,122 patents were obtained by the end of 2019. SMIC has gone all out and can only wait for the opponent to slow down and seek opportunities. Will the opponent in
give such a chance? It's really possible.
third-party consulting agency IBS reports that as the process process continues to upgrade, the R&D costs of enterprises are increasing, and the rate of increase in R&D costs per two generations is basically maintained to increase by 100%. As an industry leader, every time TSMC takes a huge amount of money under your feet.
spent money and got things done. Of course, this is the best, but the ending may not be so. Once you encounter process traps or technical watersheds, the market structure is likely to change.
In 2014, the TSMC process process was upgraded to 20nm, and the Snapdragon 810 solution was manufactured for Qualcomm . However, the kernel architecture of this solution is too focused on improving performance, which leads to severe heat from the entire machine and deep into the process trap, which ultimately creates the dark history of Qualcomm's "Fire Dragon Processor". TSMC was forced to speed up the process upgrade, and it was only after two technical iterations in 2015 and 2016 that it was rid of the shadow.
In 2015, when upgrading the 10nm process, facing the innovative EUV and traditional 193nm deep ultraviolet light source equipment, Intel finally chose to evolve conservatively, resulting in the corresponding solutions being delayed and not finally mass production was achieved in 2019. Kozaiqi, then CEO of Intel , also admitted that the company was always struggling at that time.
It is precisely because of the mistakes of TSMC and Intel that the 20nm process has also become a node that latecomers can avoid, and EUV eventually becomes a key technology to promote the semiconductor industry. As the industry leader, TSMC has won more than half of the semiconductor industry and must bear all the risks brought about by this. Once you encounter a level, SMIC can ultimately benefit as a follower.
At present, 5nm arrows are on the string and 4nm is not far away. It is difficult to predict which process node will become a new trap; in the face of the continuous emergence of innovative technologies, it is not surprising that who will become the Intel , which is a mistake. Therefore, for , making mistakes for , TSMC , is more unacceptable than slowing down, which also gains more time for SMIC to catch up.
As for whether you can finally get ashore, you can only wait for time to give the answer.
04, Conclusion
Frankly speaking, it is indeed a bit difficult to make SMIC challenge the dominant position of TSMC . This is a contest where the strength gap between various aspects is still very different. In the short term, it is difficult to find a chance of winning by SMIC . The greater possibility is that TSMC may not necessarily regard SMIC as a real challenger.
However, looking back at the development of various industries in China, many companies have experienced such tempering: Huawei in the communications industry, BYD in the automotive field, Alibaba in the cloud computing market... When they stand on the ring, they have no absolute chance of winning, but they must face the difficulties and eventually achieve different degrees of success. In this process, no one can help you, you can only help yourself.
At present, the situation facing SMIC is more steep and the opponent is stronger. However, as long as you persist and as long as you are still on the battlefield, you will always find a chance to succeed.
