In-depth review of China’s semiconductor industry, missing 30 years of gold

In-depth review of China’s semiconductor industry, missing 30 years of gold

“In the 1960s, we fully invested in two bombs and one satellite, and we gained a lot. In the 1970s, we did not engage in semiconductors, and we lost a lot because of this,” Qian Xuesen said with emotion in his later years.

In January 1950, Wang Shouwu, who taught at Purdue University in the United States, was inspired by the American Association for Science and Technology and decided to return to China to make a contribution to the newly liberated New China.

However, when the Korean War broke out this year, the Truman authorities obstructed the return of Chinese students in every possible way. Wang Shouwu had to return to his hometown to visit his elderly widowed mother. He entered Shenzhen from Hong Kong through the assistance of the Indian embassy in the United States and returned to China in a curve.

Returning to China with Wang Shouwu, there are Huang Kun from the University of Bristol, Xie Xide from the Massachusetts Institute of Technology, Xia Peisu from the University of Edinburgh, Tang Dingyuan from the University of Chicago, Huang Chang from Harvard University, and Lin Lanying from the University of Pennsylvania. .

It is these returned scientists who laid the first foundation for the new China’s semiconductor industry.

These talents cannot bring back foreign equipment, but can only bring back the knowledge in their heads. They prepare materials, build equipment, and compile textbooks by themselves. They have trained the first batch of students in the field of semiconductors in New China, and started their careers from scratch.

During this period, the gap between Chinese semiconductors and the United States, the birthplace of technology, was only 5-7 years. Driven by the demand for major projects such as “two bombs and one satellite”, the technological breakthroughs in some fields are even earlier than that of Japan and South Korea. For example, the breakthrough of planar technology is only 5 years later than Fairchild Semiconductor Noyce, and the development of the first integrated circuit is only 7 years later than that of the United States.

Some people call China’s semiconductor industry a “dream start”, but the fact is that the Chinese are gradually being separated by the United States and even Japan and South Korea in the process of marching into large-scale and ultra-large-scale integrated circuits.

Looking at the development of integrated circuits in China and the United States, Japan, South Korea and other countries on the same timeline, China’s laggard began in the 1970s, and the gap reached the largest in the 1980s. Although it began to catch up in the 1990s, it still faltered. , which is still on the way to this day.

What this article hopes to recap is, what did we miss in the 1970s and 1990s, and what was the problem? What enlightenment can it bring to today’s “nationwide core building”?

Exciting technology introduction

In May 1973, Wang Shouwu, who was the deputy director of the Institute of Semiconductors, Chinese Academy of Sciences, led a team of 13 experts to visit Japan.

More than 3 months before departure, the Fourth Machinery Department (later renamed the Ministry of Electronics Industry) held a symposium on integrated circuits. An important topic at the meeting was to point out the quality problems of the integrated circuits produced by the state-owned Dongguang Factory (also known as the 878 Factory). After investigation after the meeting, 878 Factory summed up the quality reasons into four words: dirty, empty, wounded, and leaking—that is, dirty, empty welding, scratches and air leakage.

In-depth review of China’s semiconductor industry, missing 30 years of gold

Group photo of some Tsinghua alumni from 878 Factory

At this time, Japan, which is across the sea, has initially established its own semiconductor industry system through the combination of government, industry and academia with the support of the United States. After the introduction, digestion and absorption of American technology in the 1960s, the Japanese semiconductor industry has gradually acquired the ability for secondary innovation. By 1969, Hitachi was able to independently develop and begin mass-manufacturing all-transistor color TV sets.

This is the first time that technicians from the Chinese semiconductor field have organized a delegation to visit Japan after the normalization of diplomatic relations between China and Japan in 1972. It was not easy to go abroad in the 1970s. Wang Shouwu and the experts of the delegation cherished the opportunity to visit Hitachi, Toshiba, NEC, Panasonic, Mitsubishi, Fujitsu and Sharp.

In addition to looking at design and production processes, experts also pay special attention to equipment and technical processes. From this point of view, it is found that the gap is not small. Japan has been able to mass produce MOS integrated circuits in 1972, and some companies have begun to use 3-inch wafer production lines. And the Chinese are still solving the quality problem of small-scale integrated circuits.

The rise of the semiconductor industry in Japan and later South Korea and Taiwan, China has benefited from the transfer of technology and industry from the United States. After the semiconductor technology originated from the United States took root in these countries and regions, it in turn supported the American electronics industry – and as a China, a socialist country, is naturally the object of strict prevention and control.

At that time, there was a well-known specialized organization called the “Paris Coordinating Committee”, which specifically targeted socialist countries to impose embargoes, and China was the most closely watched. On the list of four categories of embargoes, China has only one item, which is more than 500 items more than those applicable to the Soviet Union and Eastern European countries. In the 1990s, after the dissolution of “Batumi”, the Wassenaar Agreement was replaced by the N-2 approval principle, that is, any technology imported into China is at least two generations later than that of Western countries.

In this long-term blockade, the Chinese have been pondering semiconductors by themselves behind closed doors. It is not easy to develop semiconductors on your own, especially during the ten-year turmoil, when a large number of scientists were criticized and delegated, and they could only do some theoretical research secretly while cleaning the toilet.

The honeymoon period between China and Japan became an opportunity that China almost seized in the mid-1970s. The inspection in 1973 had an unexpected harvest. NEC expressed its willingness to transfer a complete set of advanced 3-inch chip production lines to China – this is a relatively recent opportunity for us to introduce advanced integrated circuit production lines. If this production line was introduced at that time, we might have started mass production of the COMS process 3 years earlier than Taiwan and 5 years earlier than Korea.

At that time, NEC’s transfer price was: US$30 million for one process technology and full-line equipment, US$40 million for two processes and equipment, and US$50 million for three processes and equipment. But at the time we were in a situation where the most we could get was $15 million.

Lack of money was also an important factor that held back China’s integrated circuit development in the 1970s. From 1966 to 1995, my country’s cumulative investment in semiconductors was only 5 billion yuan – that is to say, the total investment of the whole country in the past 30 years is less than the one-year capital investment of a large foreign company.

After Wang Shouwu returned to China, he reported the situation to Qian Xuesen, then deputy director of the Science and Technology Committee of the National Defense Science, Technology and Industry Commission, but Qian Xuesen said he was helpless. At that time, it was the period when the Gang of Four was at its most violent. Wang Zhen, the old minister of the Fourth Ministry of Machinery, was persecuted again because of the “Snail Incident”. Some people think it’s a line problem “related to national character”.

Under such a background, NEC’s complete set of equipment and technology is of course impossible to buy. The compromise solution is to purchase a single device from Japan and the United States to combine production lines with 7 domestic units, but in fact, it is impossible to scale MOS circuits. Production. It was not until 1988 that the Shanghai Radio 14 Factory and Belgian Bell established a joint venture, and China had initially built a large-scale MOS circuit production line, which took a full 15 years.

The turmoil, blockade, and lack of money made the Chinese electronics industry people of that era only able to sigh and sigh in the face of foreign advanced technology and equipment. Qian Xuesen once said with emotion in his later years:

“In the 1960s we went all-in on two bombs and one satellite, and we gained a lot. We didn’t do semiconductors in the 1970s, and we lost a lot because of it.”

National Refinery of Semiconductors

In the 1960s and 1970s, there was a “semiconductor fever” in China. Various provinces and cities have built Electronic factories one after another, and scrambled to launch integrated circuit projects. Over 40 integrated circuit factories have been built nationwide in a short period of time.

During the period of national frenzy during the Cultural Revolution, there was even a mass movement to refine semiconductors. In order to break the “cutting-edge superstition”, the newspaper also carried out a long report on the propaganda model of an old lady pulling a diffusion furnace in an alley to engage in semiconductors.

The earliest theoretical source of this mass refinement of semiconductors movement is the “electron-centered theory” put forward by Chen Boda on the eve of the Cultural Revolution, and he can be regarded as the earliest comprehension of the new technology revolution among the senior leaders of the central government. In 1965, when the Standing Committee of the Political Bureau of the Central Committee held a meeting to discuss this issue, Deng Xiaoping pointed out:

“China has a large population and a weak foundation. It may not be appropriate to develop too many new technologies. It is better to keep business as usual and be safe.”

At the Lushan Conference in 1970, Chen Boda fell from power. “Electron-centric theory” has also become the object of public criticism by the People’s Daily.

But after all, this is just an absurd episode in a special era. The main reason for the popularity of semiconductors in the 1960s and 1970s was that “Batumi” imposed a blockade on China. China’s electronics industry could only rely on its own components to support it. At that time, the price of a NAND gate circuit in China was as high as 500 yuan. Driven by profits, integrated circuit projects all over the country rushed forward.

And this kind of fiery aroused a person’s vigilance.

Wang Hongjin, the director of Jiangnan Radio Equipment Factory (also known as the 742 Factory), is an old revolutionary who participated in the War of Liberation and Resist U.S. Aggression and Aid Korea. He has studied radio communication technology in the army. Wang Hongjin thought that the market is so big that the electronics factories all over the country will definitely fight. After much deliberation, he decided to abandon the red-hot integrated circuit production and focus on discrete devices instead. This transformation has also made Jiangnan Radio Equipment Factory a leading company in discrete devices, gaining a firm foothold in the market competition.

In-depth review of China’s semiconductor industry, missing 30 years of gold

In 1969, Jiangnan Radio Factory moved into the Dawangji factory area

Wang Hongjin’s judgment is undoubtedly wise now. Although a large number of integrated circuit projects were launched in China, most of them were scattered and inefficient production methods. A large number of semiconductor factories are also scattered, small, independent and weak. In July 1977, Comrade Deng Xiaoping, who had just returned to be in charge of science education, invited 30 representatives from the scientific and technological circles to hold a symposium for science and education workers in the Great Hall of the People. Wang Shouwu said:

“There are more than 600 semiconductor production plants in the country, and the total amount of integrated circuits produced in one year is only one-tenth of the monthly output of a 2,000-person factory in Japan.”

Also in March of this year, Jiangnan Radio Equipment Factory faced the second major decision. Wang Hongjin, with more than 10 people including chief engineer Che Yunhong, went from Wuxi to Beijing to negotiate with Japan’s Toshiba. Just a few months ago, the country decided to introduce color picture tube production lines and integrated circuit production lines from Japan. In order to settle down this production line, electronic factories all over the world “show their magical powers”, and Wang Hongjin naturally does not dare to slack off.

After layers of investigation and demonstration, this production line finally ended up in Jiangnan Radio Equipment Factory. But it didn’t take long before the question came – just to implement the standard for a product, the factory was full of quarrels, whether to stick to Japanese standards to regulate production, or to guide production based on user needs?

This problem does not seem to be a problem now, but it is a huge ideological challenge for enterprises in the early stage of reform and opening up. Finally, under the insistence of a group of people, 742 Factory chose to be oriented by user needs. It is the simple “market awareness” in the context of this planning system that once again makes Jiangnan Radio Equipment Factory stand out.

At this time, there has been a climax of importing foreign equipment in the whole country. The electronics factories that were competing to be built in the past are now competing to invest in the introduction of foreign equipment. From 1981 to 1985 alone, 33 units across the country introduced production equipment to varying degrees, with a cumulative investment of 1.3 billion. However, because most of the imported equipment is obsolete, and there is no supporting technology and management, only a few lines can actually be put into production in the end.

The background at that time was that the state reduced direct investment in the electronics industry and encouraged electronics factories to find their own way out in the market. The electronics factories that were recently driven out of the sea lacked scientific research capabilities and industrialization experience, and blindly followed suit. Enterprises that have successfully introduced equipment have no ability to absorb and digest technology, let alone secondary innovation.

Five years after the Toshiba production line was established, a Japanese person in charge came to China to understand the situation and found that the production process had not improved at all compared with 5 years ago. The Japanese who had some concerns about transferring technology also went back with confidence.

In the early stage of the development of China’s semiconductor industry, relying on a group of overseas returnees, Soviet consultants and domestic technicians to break through various technologies, they did not fall behind much in scientific research. However, scientific research results are not equal to industrialization capabilities, nor can they represent the capabilities of mass production and mass production.

In contrast, Intel in the United States has been committed to breaking the barriers of R&D and production since its inception. The development of semiconductors in Japan and South Korea is also the result of a high degree of integration of production, education and research. Due to the early system, the Chinese often use semiconductors as a breakthrough in scientific research projects, but do not cultivate and develop them as an industry.

Chinese semiconductors, which did not lose on the starting line of scientific research, were left behind in the industrialization competition.

Unable to catch up with Moore’s Law

In order to solve the problem that “scientific research results cannot be transformed into productive forces”, the relevant leading departments moved a part of the scientific research force from Yongchuan’s 24 research institutes to Wuxi, and formed a production-research consortium with Jiangnan Radio Equipment Factory, which later developed into Huajing Electronics Group Corporation. However, from the perspective of Huajing’s repeated introduction of technology from abroad, the connection between R&D and production is not satisfactory.

However, this occasion did not cultivate a group of core-making talents for China. Huajing later became the Huangpu Military Academy of China’s integrated circuit industry, and cultivated no less than 500 key talents for relevant government departments, SMIC, Huahong Grace, Huali Microelectronics, Changdian Technology, Tongfu Microelectronics and other enterprises .

On the basis of Huajing, the State Planning Commission and the Ministry of Electronics Industry proposed the “908 Project” in 1990, with the goal of improving semiconductor technology to 1 micron during the “Eighth Five-Year Plan” period. The total cost of the project is 2 billion yuan – of which 1.5 billion yuan will be used by Huajing Company to build a fab with a monthly output of 12,000 wafers, and the other 500 million yuan will be invested in the construction and design centers of 9 integrated circuit companies. At the time, this project was highly expected by the whole country, and it was a battle determined to narrow the gap with the world’s advanced level.

At this time, the global electronics industry has fully entered the stage of VLSI development, and the power of Moore’s Law is prominent. The so-called Moore’s Law is a well-known law of chip industry development proposed by Gordon Moore, one of Intel’s founders, in 1965, that is, the number of circuits integrated on an integrated circuit chip doubles every 18 months.

This leads to the anti-Moore’s law – selling the same number of products with the same performance as 18 months ago, the turnover will drop by half. Anti-Moore’s Law forces all chip companies to catch up with Moore’s Law, and any company that cannot keep up with technology will be eliminated.

Under the impetus of this law, the world’s chip pattern is surging, and the competition between chip companies and even countries and regions can be described as tragic.

In 1981, Japan’s Panasonic launched a competing memory chip for Intel, causing the price of Intel Chips to drop from $28 to $6. The following year, Japan’s Toshiba invested 34 billion yen, and a team of 1,500 people led by Kawanishi started to implement the “W Plan”. Three years later, 1MB dynamic random access memory was mass-produced, and the “price is always 10% lower” competition was launched. strategy to bring Intel to the brink of bankruptcy.

In 1985, under the leadership of its outstanding manager Grove, Intel made a “strategic turn”, abandoned the memory chip business it started, and focused on personal computer CPUs, and began an invincible journey on another track.

In the same year that China decided to implement the “908 Project”, Intel’s 386 chip was equipped with Microsoft’s Windows 3.0 system, which completely changed computer technology and opened the “Wintel era” that monopolized the desktop for more than 30 years. Also in this year, South Korea’s Samsung developed the world’s earliest 256MB dynamic memory, officially announcing that it surpassed Japanese technology. Three years ago, Zhang Zhongmou, the No. 3 figure of Texas Instruments, returned to Taiwan, China, and founded a company specializing in OEM chip production, called TSMC.

The world semiconductor situation is changing rapidly, but our 908 project has been “in preparation”. It took 2 years to approve the cost alone, 3 years to introduce the 0.9-micron production line from Lucent in the United States, and 2 years to build the factory – it took 7 years from project establishment to production.

7 years is equivalent to 4.6 Moore’s Law cycles, which means that the production is nearly 5 generations behind the mainstream technology. In the year when it was put into production, Huajing lost 240 million yuan, and the huge investment went to waste.

A more serious consequence is that the failure of the “Eighth Five-Year Plan” for integrated circuits has caused Chinese chips to miss another five years. When Huajing Company was still unable to mass-produce the 0.8-micron card point, the world’s mainstream manufacturing process has developed to 0.18-micron.

Plan and market two hands

The “908 Project” is mainly financed by loans from China Construction Bank.

The huge loan interest of commercial banks has increased the burden on enterprises, and Huajing has changed from a profitable unit to a loss-making unit. Due to serious bad debts, Huajing failed to even try to convert debt to equity, and was eventually taken over by China Resources Group, a Chinese-funded enterprise in Hong Kong, and changed its name to “China Resources Microelectronics”. When talking about the matter, Zhu Yiwei, an older generation semiconductor expert, said:

“It is not feasible to rely solely on bank loans with higher interest rates to build chip projects.”

Another person trapped by loan interest is Shougang R&D. In 1991, with the determination that “Shougang will not be named steel in the future”, Shougang crossed into the chip industry and established a joint venture with Japan’s NEC. However, the total amount of capital invested by China and Japan is less than 1/3 of the total investment amount, and the rest of the funds basically come from domestic and overseas bank loans, burdening them with a heavy interest burden.

In addition to the funds, Shougang NEC also has other problems – the technology is completely provided by NEC, the main customer is also NEC, and Shougang NEC is only producing against Japanese drawings.

This “two-headed-out” model leaves companies with little ability to resist risks. When the semiconductor crisis in 2001 hit NEC hard, Shougang NEC was immediately in trouble. In 2004, Shougang announced its complete withdrawal from the chip industry.

In-depth review of China’s semiconductor industry, missing 30 years of gold

The old style of Shougang Ridian

What is intriguing is that in the year before exiting the chip industry, the real estate business developed into one of the core businesses of Shougang, and even became one of the top 100 real estate development companies in China in the future.

This is not an isolated case in Beijing. Most of the semiconductor giants in Beijing back then spun off their semiconductor business and instead invested their funds in the rising real estate market. Only BOE still persisted alone on the road of research and development. Despite repeated defeats, it was determined not to engage in real estate until it finally became a leading enterprise in LCD panels in China.

In the eyes of the stockholders at the time, BOE was a company that “did not act according to common sense”. It is only natural for an enterprise to pursue profit, and the board of directors also needs to explain to shareholders. When making decisions on the development direction, there are often more “rational” and “realistic” considerations. A capital-intensive, technology-intensive, and long-returning industry like semiconductor integrated circuits cannot be developed solely by planned scientific research, or by simply relying on the “invisible hand” of the market.

The two hands of planning and market need to function at the same time, and the two hands cannot fight with each other.

By 1994, the backwardness of chips in mainland China could be described as “shocking”. This year’s production and sales only accounted for 0.3% and 0.2% of the world market share respectively, and the technology level was more than 15 years behind developed countries.

In this context, the country launched the “909 Project”, which is the second time that the Chinese have attacked the “Core” highland after the setback of the 908 Project. In his later memoirs, Hu Qili, then the Minister of Electronics Industry, wrote down the rhetoric of “success, not failure” at that time, but then wrote:

“Now that I think about it, my estimates of the storms and dangers that I would encounter at that time were far from enough.”

Around the “909 Project”, electronic enterprises such as Shanghai Hongri International, Shanghai Huahong NEC, and Shanghai Huahong International were successively established in Shanghai in 1997. Among them, Hua Hong NEC learned the lessons of the “908 Project” and completed the trial production in less than two years. In 2000, it achieved a good performance of 3 billion sales. But the good times did not last long, and the global semiconductor crisis was encountered in the second year. As a new enterprise, Hua Hong NEC was not spared in this crisis, with a loss of 1.348 billion yuan that year.

Due to the lack of understanding of the integrated circuit industry, in the first few years of Hua Hong NEC’s establishment, “profit or loss” became the main criterion for evaluating its success. When it suffered a loss in 2001, the voices of criticism came one after another, and the media also pointed out:

“You can’t make chips just by throwing money at it.”

Three years later, although Hua Hong NEC’s performance has returned to stability, it has not received national financial support for more than ten years since then.

Spending money may not necessarily make chips, but making chips must cost money.

In 2000, under the impetus of four domestic semiconductor academicians, the Beijing Municipal Government was determined to build the northern microelectronics industry base. The following year also due to the global semiconductor crisis, overseas fundraising was in trouble, and Chuangxun and Huaxia projects were aborted.

But in fact, some far-sighted people will choose to invest and build factories during the downturn of the industry. The cost of building a factory during the low tide period is relatively low, and it is likely to catch up with the next round of high tide after completion. Unfortunately, it is difficult for most investors and operators to make such decisions. Those who really dare to invest their capital in a counter-cyclical situation are often inseparable from the guidance and firm support of the government.

In 1983, South Korea’s Samsung decided to fully enter the chip, but encountered an extremely dangerous start. When Samsung introduced 64kb memory chips, the US-Japan memory chip price war was in full swing, with memory prices dropping from $4 to 30 cents. Samsung, which was first involved in chips, was involved in the torrent of price wars and lost all its equity capital in just two years. Founder Li Bingzhe has lingering fears afterwards:

“For this project, Samsung bet everything.”

According to Li Bingzhe’s recollection, when every department came to report to him at that time, without exception, they cried out that they were about to be unable to hold on, and persuaded him to withdraw from the integrated circuit industry as soon as possible to leave a way out for himself.

In the 1960s, Li Bingzhe bought 200 acres of land in the suburbs of Suwon to build an electronic industrial base, and now it has developed into the headquarters of Samsung Suwon.

At the critical moment, it still relies on the South Korean government to take action, not hesitate to use Japan’s war reparations, invest 346 million US dollars, and at the same time drive 2 billion US dollars of private funds to Samsung trays. It is under such blood transfusion that Samsung has the “counter-cyclical investment” that later made it famous, and finally made it to the United States and Japan to sign the “Semiconductor Agreement”, a step forward to fill the gap of the Japanese.

Riding the wind and waves sometimes

The 1970s and 1990s when China’s semiconductor industry was hesitant was a major turning point in the global semiconductor industry’s competition.

From the introduction of DRAM dynamic memory by Intel in 1971, which marked the beginning of the birth of large-scale integrated circuits, the following 30 years have been the golden age of global large-scale integrated circuit industrialization. Finally formed today’s world chip pattern.

From 1976 to 1979, the Japanese government led the famous “VLSI Project”. Led by the Ministry of International Trade and Industry of Japan, Hitachi, Mitsubishi, Fujitsu, Toshiba, and NEC, five companies as the main body, established the “VLSL Technology Research Group”, focusing on superior talents to cooperate in tackling key problems and breaking the technical barriers of enterprises. The “movement” of the target was astonishingly effective, and more than a thousand patents were obtained. The level of Japan’s integrated circuit technology was greatly improved, laying the foundation for Japan’s memory chips to surpass the United States in the 1980s.

In 1986, after encountering the containment of Japanese integrated circuits, the U.S. government personally confronted it and released the report “Strategic Industries in Crisis”, which led Japan to sign a series of regulations restricting the development of its integrated circuits. Provide various support to US semiconductor companies.

In the same year, the South Korean government joined forces with Samsung, LG, Hyundai and six South Korean universities to develop 4MB dynamic memory chips as a national key project. It invested 110 million US dollars in three years, and the government took the initiative to undertake 57% of the funds. Chips beyond Japan laid the foundation.

In 1987, TSMC was established. The Taiwan authorities set up a special development fund and invested US$100 million to account for 48.3% of the shares. Politicians even went out to raise funds for TSMC in person. In addition, various concessions such as venue and tax are also given. The intensity was so great that there was even a spectacle that TSMC’s after-tax profit was higher than its pre-tax profit.

In the integration of the global industrial chain, enterprises have developed core technical capabilities of independent innovation, and under the overall planning of the government, they have made continuous, stable and constructive capital investment; in the fierce competition with the United States, the birthplace of technology, incubate and A competitive enterprise that grows up; in the integration of government, production and research, a leader and a mature team of technology and management have been tempered…

All the above conditions were not available in China in the early days of reform and opening up. China’s integrated circuit industry has been swaying and tangled between planning and the market for more than 30 years. When it caught up with it in the 1990s, it was found that it was still shackled and floundering.

Fortunately, around 2000, a group of new-generation chip companies, represented by SMIC, Huawei Hisilicon, Spreadtrum, and Zhongxing Microelectronics, were born. Some of them have learned from the development lessons of China’s integrated circuit industry in the last century, and gradually matured in China. It took root in the industrialized soil, and after more than 20 years of difficult growth, it finally came to fruition.

Their rise and breakthrough will be another story of China’s chip industry.

The Links:   G150XG01 V3 LJ64EU34