(in late August, Hongxin project construction site in Wuhan airport economic and Technological Development Zone. Picture / peoples vision)
Reporter / Du Wei
Issued in issue 969 of October 26, 2020
On the evening of October 4, SMIC, the largest and most cutting-edge chip foundry enterprise in the mainland, issued an announcement saying that it was confirmed that it was subject to us export control, and that some US equipment, accessories and raw materials exported to SMIC international must apply for export license before continuing to supply.
In fact, as early as 2014, with the promulgation of the outline for promoting the development of the national integrated circuit industry and the launch of the national integrated circuit industry investment fund (hereinafter referred to as the big fund), the chip industry development boom has been set off all over the country, and a project investment is often tens of billion yuan or even 100 billion yuan. In the past year or more, semiconductor projects with investment targets of 10 billion yuan have been suspended, causing a wave of unfinished business.
On October 20, at a regular press conference of the national development and Reform Commission, Meng Wei, spokesman of the national development and Reform Commission, said that the development order of integrated circuit industry should be orderly guided and standardized in accordance with the development principle of main body concentration and Regional Agglomeration; the prevention mechanism should be established to guide local governments to strengthen the risk awareness of major project construction, and inform the local authorities of major losses or risks Responsibility.
Where is the neck stuck
The birth of a chip has to go through three steps: design, manufacture and package. At present, the gap between the domestic chip design and the world advanced level in the field of design and packaging is constantly narrowing, and the neck sticking link mainly lies in manufacturing. Huawei is a typical example. Taking Kirin 9000 chip as an example, because it only has the layout of design and not involved in manufacturing, it can only find TSMC, which has the worlds leading 5-nanometer process technology, as its OEM.
The so-called 5nm refers to the distance between the source and drain on the transistor, the smaller the value, the more transistors can be arranged on the chip. Currently, TSMC is developing a 3-nanometer process, which is expected to be put into trial production in the second half of 2021. SMIC has achieved mass production of 14 nm process in 2019, which is at least two generations behind TSMC.
Even more fatal is the lack of manufacturing equipment and materials. When making chips, the raw material silicon should be purified first, and then cut into silicon wafers with uniform thickness and no more than 1 mm, which is called wafer, which is equivalent to the foundation of the chip. The larger the wafer, the more chips can be packaged, the lower the cost and the higher the production process requirements.
Wei Shaojun, vice president of the China Semiconductor Industry Association and director of the Institute of microelectronics at Tsinghua University, told China News Weekly that at present, large silicon wafers with purity of 99.9999% or higher required for IC Processing in China basically rely on imports. Although the photoresist composed of hundreds of raw materials and complicated processing technology is not insurmountable technically, it has been developed due to domestic research and development Bu Wan also relies mainly on Japanese and American enterprises.
Known as the Pearl on the crown of semiconductor industry, the importance of lithography is well known. At present, the most advanced lithography machine is the extremely ultraviolet (EUV) lithography machine produced by ASML company in the Netherlands, with a price of more than 100 million US dollars. At present, the domestic lithography technology level is 90 nm. Shanghai microelectronic equipment (Group) Co., Ltd., a leading semiconductor equipment manufacturer in China, will deliver the first domestic immersion lithography machine with 28 nm process from 2021 to 2022.
Due to the mismatch of industrial structure and demand, China mainly relies on imports for core chips such as processor / memory. On the surface, domestic enterprises have made some progress in recent years, but in fact, the progress is very fragile, because more than 90% of the industrial software MES for production equipment, raw materials and production process management have to be imported.
Even in the field of design, Chinas chip industry has weaknesses. EDA, the automation software for chip design, is controlled by two American companies and a subsidiary of Siemens in Germany. In addition, 90% of the worlds smart phones and tablet PC chips use arm architecture. Arm and Intel x86 architecture, which occupy the dominant position in the computer, are known as the two major architectures of the current chip. But on September 14, NVIDIA, an American chip giant, announced that it would buy arm, a well-known chip design and architecture company of Softbank, for $40 billion. In the outside world, this may cause the U.S. government to restrict the use of arm architecture in China.
(on October 14, visitors watch the chips produced by SMIC international OEM at the China International Semiconductor Expo 2020. Figure / IC)
The end of the chip industry
In 2014, the national large fund was launched, and the first phase raised more than 130 billion yuan. The big fund is an industrial investment fund set up by the central finance, China Development Finance and China Mobile. It mainly invests in leading enterprises in industries, such as SMIC international and Changjiang storage. It also aims to leverage more social capital injection. In Wei Shaojuns view, to some extent, large funds are more like funds set up for the expansion and listing of enterprises.
By the beginning of October this year, there were more than 50000 chip related enterprises in China, and 12740 chip companies were newly established this year.
On January 20 this year, a brand new ASML high-end lithography machine, which has not yet been put into operation and is called the only one in China to produce 7-nanometer chips by the government of Dongxihu District of Wuhan City, was mortgaged to the Dongxihu branch of Wuhan Rural Commercial Bank with 580 million yuan as the mortgagor, and the mortgagor is Wuhan Hongxin Semiconductor Manufacturing Co., Ltd. The company, which was established in November 2017, has attracted much attention because it claims to have advanced production processes of 14nm and 7Nm, and plans to invest a total of 128 billion yuan in two phases. In 2018 and 2019, Hongxin was successively selected as major projects in Hubei Province and ranked first in the 2020 municipal major project plan issued by Wuhan development and Reform Commission. In 2019, Hongxin also invited Jiang Shangyi, former chief operating officer of TSMC, to be the CEO of the company.
However, on July 30 this year, the Dongxihu District Government of Wuhan City, which introduced Hongxin, pointed out in the economic operation analysis in the investment and construction field of Dongxihu District in the first half of the year published on its official website that Hongxin project has a large fund gap and is faced with the risk of project stagnation caused by the rupture of capital chain at any time. At present, Wuhan Hongxin has been removed from major projects in Hubei Province.
In Chengdu high tech Zone of Sichuan Province, grid core (Chengdu) integrated circuit manufacturing Co., Ltd., which was jointly established in 2017 with a planned investment of US $9.053 billion by us chip manufacturer grofande and Chengdu municipal government, has been closed. The company once said it would build the largest 12 inch wafer plant in China.
In June this year, the star company of semiconductor industry, Nanjing decama, entered into bankruptcy liquidation and asset transfer procedures due to lack of funds. The case of Decoma has attracted much attention. Li Ruiwei, its founder, is known as a speculator for his participation in the establishment of semiconductor companies in Huaian, Nanjing and Ningbo.
However, according to media reports, by October 2019, only 4.6 billion yuan has been invested in the project, which has basically run aground. China Newsweek learned that the local government has set up a work promotion group of Dehuai semiconductor to take charge of the overall stability work. After the establishment of Dehuai, Li Rui returned to Nanjing for the conflict with Xia Shaozeng and founded Nanjing decama. The total investment target of phase I is expected to reach 5 billion to 6 billion yuan, and plans to produce power management chips and other analog chips. However, more than one year after the project was started in January 2017, the capital chain has broken. By February 2019, the investment raised by decama was only 250 million yuan. After the Nanjing project was suspended, Li Rui took more than a dozen old departments and registered in Ningbo Chengxing (Ningbo) semiconductor company, a chip design company that does not need to spend a lot of money to build a factory.
Sorting out these unfinished companies or projects in China, most of them have a common feature: the project side has little or even zero cost investment.
Gu Wenjun, an expert in semiconductor industry and chief analyst of xinmou research, told China News Weekly that a very important problem in the industry now is that the government has invested money to set up enterprises in person, and there is a lack of awe for the semiconductor industry. What local governments should really do is to build a platform and build a good business environment. Gu Wenjun also wrote that some enterprises which are light on assets (or even have no assets) go to the local government for landing. If they have no products, the local government will first receive subsidies of several or even more than a billion yuan.
Shi Qiang, director of the integrated circuit manufacturing research office of the Institute of integrated circuits, saidI think tank, said that many semiconductor projects are speculative. In order to take the lead, some projects have been launched in a low-key way, even with a mysterious color. Due to the lack of actual investment and the lack of deep binding with the local government, it can be said that the project side can just walk away. Some projects themselves are intended to defraud money.
According to a senior person in the semiconductor industry, the semiconductor industry has the characteristics of capital intensive, technology intensive and talent intensive. In many places, these conditions are not met, and the huge amount of investment claimed can not be reached. Therefore, only a few projects can be carried out, and other projects will be completed.
Low level repeated construction and false fire Market
According to the investment and construction agreement signed by Li Ruiwei and Nanjing Economic and Technological Development Zone in 2015, what Nanjing dekoma initially wanted to build is an industrial park of Image Sensor CIS, which is said to fill the gap of CIS industry in China. CIS chips can be used for mobile phone cameras, security monitoring and vehicle mounted mobile cameras. According to the requirements of pixels and resolution, CIS chips can be divided into low-end and high-end. The idea of Li Ruiwei and Nanjing is to start with the common, medium and low-end, and then gradually upgrade. After the restart of the Nanjing decama project in 2017, both sides felt that the difficulty of CIS chips was still high, so they decided to make analog chips with relatively lower process requirements first, and CISs grand plan was first realized in Dehuai semiconductor.
In June 2017, the main plant of Dehuai semiconductor phase I 12 inch wafer plant was capped. In the same year, Xia Shao introduced the technology of American Anson. But according to Caixin, by the end of 2019, the total sales volume of Dehuai products was only 250 million yuan. Around 2016, Cao Yun, who was still working for the industry giant STM semiconductor at that time, had some knowledge of the decamer project, which was unbelievable at that time. He believes that there are already some domestic enterprises like Shanghai Huahong doing CIS, and some OEM factories also have mature projects and technologies. Without much professional foundation, decama will enter the CIS without much professional foundation, so it will not be competitive. In terms of national strategic layout, it is not necessary.
Cao Huashi is the founder of eggplant (Shanghai) Management Consulting Co., Ltd. as the program planner of chip unveiling, he has many contacts with the industry. In her view, there are too many enterprises doing power devices. Relatively speaking, the consumption is relatively large, and the threshold is not high. Many start-ups will take this product to practice, or fill in the capacity first. In the whole country, low-level repeated construction cases appear frequently.
Now there are the third generation semiconductor projects in China. At present, most chips use silicon as the main raw material, and the second generation semiconductor material is gallium arsenide. The third generation of semiconductors, the so-called compound semiconductors, use silicon carbide and gallium nitride as the main raw materials. Strictly speaking, this should not be called the third generation semiconductor, but the third kind of semiconductor. The third kind of semiconductor is not more advanced than the previous two generations, nor can it replace the former two generations. However, each has its own uses. Many project parties deceive the government with the concept of the third generation semiconductor.
Among the semiconductor chip markets mentioned above, senior people account for less than 4% of the global semiconductor market, with silicon as the third major raw material. The third kind of semiconductor has the characteristics of high pressure resistance, high temperature resistance and high speed. It can be used in 5g, 6G network communication chips, blue laser lighting devices, and high-voltage resistant devices in the new infrastructure. However, its process requirements are not high. The highest precision only needs 250 nm, which can be achieved 30 years ago when silicon chips were prepared. The precision requirements are rough.. However, the core problem is that raw materials are relatively difficult to prepare, raw materials need to be 100% imported. After decades of development, the United States and Japan have just started mass production, which is not so easy. At present, there are no manufacturers in China, they are all speculating on the concept. Moreover, the price of the third semiconductor product is higher than that of the other two semiconductors. This also determines that it is a niche market and ordinary people may not be able to afford it.
Wei Shaojun once pointed out that in the case of insufficient capacity of advanced process nodes, the planned capacity of the domestic semiconductor industry is mainly concentrated in the 40 nm ~ 90 nm process, and it is expected that some nodes may have excess capacity after completion. Another direct consequence of repeated construction is that investment will be dispersed.
The fast-moving chip industry is also pushing up the market value of entrants. According to Cao Huashis observation, among the more than 3000 chip design enterprises in China, few enterprises have their own intellectual property rights and design high-end chips according to market demand. However, after designing a chip, many enterprises will take advantage of the convenience of the listing process and threshold of the scientific and technological innovation board and rush to the market. In this industry, it is not difficult for enterprises to sell more than 100 million yuan annually.
According to the data of CCID consultants, from January to August this year, the IPO fund-raising scale of Chinese semiconductor enterprises reached 66.174 billion yuan, 5.93 times of the whole year of 2019 and 42.47 times of the whole year of 2018. The price earnings ratio of the semiconductor enterprises on the science and technology innovation board, that is, the ratio of market value to profit is more than 100 times, which means that the semiconductor enterprises have obtained a high valuation.
Shen Yinlong sighed to reporters that when an emerging industry comes up, it is easy for everyone to rush in a short time. Local governments rub their hands and investors are eager to try, which is a kind of blindness. This is a malpractice in the development of our industry and forms a kind of virtual fever..
Where is the way to solve the problem of neck jam?
Shen Yinlong told China News Weekly that when we introduced the decamer project, when we evaluated and investigated the technologies and teams involved in the project, we also communicated with relevant experts from local universities and scientific research institutes, but we are not professional and technical personnel after all.. He hopes that in the future, when semiconductor projects are introduced in various regions, experts from higher-level authorities such as the national or provincial development and Reform Commission can organize experts to conduct preliminary demonstration on the feasibility of the project.
Wei Shaojun called the crazy launch of semiconductor projects around the country as investment promotion driven, project driven, rather than scientific driven. When you ask a lot of local governments who are the target customers of their semiconductor projects, which segments they are making things for, and what the competition strategy is, they cant say.. In his opinion, there is a lack of top-level design for semiconductor projects in various regions. At present, the guiding mechanism of the national development and Reform Commission is the after the fact guidance for projects to be launched, and the gateway should be moved forward.
In 2019, the second phase of the national large fund was established, with 200 billion yuan invested in the chip market. Ding Wenwu, President of the big fund, previously said that the second phase will continue to support enterprises with layout in etcher, film equipment, test equipment and cleaning equipment, and accelerate the investment layout of core equipment and key parts such as lithography machine, chemical mechanical grinding equipment, etc.
In the view of the senior people in the chip industry mentioned above, large funds should pay attention to the investment in equipment, raw materials and industrial software earlier, and take these as the investment focus. In addition, the total investment of large funds is not enough. Its not too much to add a zero to the end of the number. Xie Zhifeng, associate professor of Massachusetts Institute of technology, winner of Intels highest achievement award and one of the founders of SMIC, told China News Weekly that in the first five years of the big fund, the national average annual investment is 5 billion US dollars, while the annual R & D investment of Intel alone is 12 billion US dollars.
In the United States, for example, R & D investment in the chip industry accounts for 17% of the sales revenue, which is twice that of other countries. In contrast, the overall R & D investment of chip enterprises in mainland China is insufficient, accounting for only 8% of sales revenue on average. Moreover, according to Moores law, integrated circuit products are updated every 18 months or so, while China usually invests in accordance with the pace of five-year planning, which results in the disconnection between scientific research investment and industrial development, resulting in R & D investment sometimes not at all. Wei Shaojun believes that in the case of good enterprise scale and profitability, the national level should increase investment in technology research and development, set up special R & D funds, change the investment rhythm, and realize innovation driven.
Another problem is talent. At present, there are about 300000 IC talents in China. From leading figures, to middle and high-level management talents, to R & D talents and engineers, there is an overall shortage. Xie Zhifeng said. Ren Zhengfei, President of Huawei, also visited Shanghai Jiaotong University, Fudan University and Southeast University from July 29 to 31 this year, stressing the need to strengthen cooperation between industry, University and research.
In August this year, the document several policies for promoting the high-quality development of integrated circuit industry and software industry in the new period issued by the State Council mentioned that it is necessary to continuously explore and construct a new national system for tackling key core technologies under the conditions of socialist market economy. Wei Shaojun believes that in the implementation of science and technology projects in the past, there was too much government involvement in scientific research objectives and final decision-making, and some methods such as compiling guidelines and project evaluation were inefficient. In the future, in the exploration of the new national system, the technical research organization should have professional and technical background, and not only manage the results, but also provide coordination and support in the whole process of tackling key problems.
Looking back on the development of Chinas integrated circuit industry for more than 60 years, Wei Shaojun said that before the 1980s, China was forced to innovate independently under the condition of being completely blocked. After the 1990s, China was mainly introduced. In the first decade of the 21st century, the development of IC mainly depends on the market mechanism, and the role of the government is somewhat lacking. In the past 10 years, the role of the government has recovered, and it has begun to work on original innovation.