On December 29, the 2022 China (Shenzhen) Integrated Circuit Summit (ICS2022), hosted by the Shenzhen Municipal People's Government, the China Semiconductor Industry Association Integrated Circuit Design Branch, and the "Nuclear High-Tech" National Major Science and Technology Project Overall Expert Group, was successfully held at the Grand Skylight International Hotel in Pingshan, Shenzhen.
Professor Wei Shaojun, Chairman of the Integrated Circuit Design Branch of the China Semiconductor Industry Association and Chief Technical Engineer of the Nuclear High-Tech National Major Science and Technology Project, gave a report titled "Recognize the Situation and Strengthen Confidence" and shared the "Five Major Predictions" for the future of China's semiconductor industry.
Prediction 1: The United States’ containment of my country’s semiconductor industry will gradually move towards one-way “semi-decoupling”. One-way means that the United States wants to unilaterally decouple from China, but we do not want to; "semi-decoupling" means that the United States will not give up the Chinese market for the time being, so there is partial decoupling.
In recent years, semiconductors have become the focus of the competitive game between China and the United States. The United States also continues to "weaponize" semiconductor technology and continuously suppress and restrict the development of China's semiconductor industry through a series of strict export control measures. In this process, the National Security Council (NSC) and some American think tanks played a role in fueling the flames.
For example, American AI strategist and Google founder Schmidt once said that the United States must ensure that it is at least two generations ahead of China's semiconductors and must jointly compete with Japan, South Korea and Europe. Matheny, the head of U.S. China policy, also believes that semiconductors are the real basis for ’s leading position in artificial intelligence.
On October 7 this year, the United States introduced a series of export control policies towards China. The blueprint for this policy came from the Center for Security and Information (CSET) founded by Matheny, the head of US technology for China.
The United States also introduced the " Chip and Science Act " this year. In addition to providing high subsidies to encourage semiconductor manufacturers to build factories in the United States, it also legally requires that any company that receives funding from the US government cannot invest in new advanced production capacity in mainland China. At the same time, the United States has also actively attracted Taiwan, Japan, and South Korea to establish the so-called Quadruple Chip Alliance Chip4, and has established an emerging technology committee with the European Union. The goal of these actions is to exclude China from its supply chain.
Professor Wei Shaojun said, "My prediction is that next year the United States will increase its efforts to contain China's semiconductor industry. Will this turn around? I personally don't have high hopes."
Prediction 2: The new central government will re-examine the semiconductor development ideas, formulate new development strategies and measures, and invest more resources. There is a high probability that it will adopt a different organizational form from the past. Scientific, comprehensive, systematic, sustained and vigorous will be the keywords.
According to Professor Wei Shaojun, there are many deep-seated reasons why the United States has continued to suppress China in recent years, the most critical of which are two major factors:
1. China has become one of the global powers. Because of this, when China is already a global power, it is impossible for China to become a vassal of other countries, nor to follow other countries. It must have its own independent development ideas.
However, Professor Wei Shaojun also pointed out that although China has become one of the global powers in terms of economic size, there is still room for further integration with the world. For example, although China's merchandise trade accounts for 11% of the world, its service trade accounts for only 6% of the world; there are 110 global Fortune 500 companies in China, but the proportion of overseas revenue is less than 20% on average; China's financial system ranks among the top three in the world, but foreign company participation is less than 6%; China's R&D expenditure ranks second in the world, but intellectual property imports are 6 times more than exports...
2. In the development process of China's electronic information industry, integrated circuits have become one of the few shortcomings of China.
From the comparison chart of China, the United States, Japan, South Korea, and India in the fields of Internet, mobile Internet , mobile communications, artificial intelligence, autonomous driving, quantum technology, and semiconductors made by Professor Wei Shaojun, we can see that China is not a "smiling face" only in the fields of quantum computing and semiconductors, while the United States is not a "smiling face" in the fields of mobile networks and quantum communications . In contrast, other countries are far behind China and the United States in many areas.
"If you remove quantum technology, China and the United States each have a weakness. The United States' weakness is in mobile networks, and China is in semiconductors. In this regard, the competitive strategy adopted by the United States is very simple. It directly suppresses China's leading companies in the field of mobile communications, Huawei and ZTE . The goal is to suppress China's advantages in this area, and at the same time contain China's semiconductors, making the United States' own advantages more obvious." Wei Shaojun said.
In order to promote the development of China's semiconductor industry, the State Council issued the "Integrated Circuit Industry Development and Promotion Outline" in 2014. In the past ten years, China's integrated circuit industry has developed very fast, with an average annual compound growth rate of nearly 20%, which is several times the global average for the same period. It has achieved many important successes, but also exposed some problems.
Professor Wei Shaojun said, "An industrial plan cannot remain unchanged for ten years. It will definitely need to be re-examined. In addition to some problems exposed in it, intermediate adjustments are inevitable. Re-examine the development ideas of semiconductors and formulate new development strategies and measures. This new development strategy and measures Implementation must be extended to 2035 or even longer, and more resources will be invested. "
It is worth mentioning that foreign media previously reported that China will invest 1 trillion yuan in the semiconductor field in the next five years, mainly in subsidies for the purchase of domestic semiconductor equipment. In response to the rumor, Professor Wei Shaojun said, "I have never heard of this. I have also verified that no one said that there is an investment of 1 trillion yuan. This may be based on rumors, or someone may be deliberately testing China. Because this news comes from foreign media."
Professor Wei Shaojun also emphasized, " I believe that with the new development strategy and measures, if there is only 1 trillion yuan in the future, it is possible We are a bit underestimated. The subsequent development of will definitely not develop in the form of 1 trillion, but in the form of a more important national strategy to develop integrated circuits. It’s hard to say how much resources will be invested in at the moment. It may be appropriate to allocate funds and spend the resources as needed based on the actual situation. Therefore, I chose a few words, scientific, comprehensive, systematic, sustained and vigorous as keywords. "
Professor Wei Shaojun further explained: "There was a passage in the 27th meeting of the Central Commission for Comprehensive Deepening of Reforms, which mentioned that it is necessary to 'focus on the research and development of key technologies with first-mover advantages and basic cutting-edge technologies that will lead future development.' There is no mention of the issue of making up for shortcomings, because just making up for shortcomings is a strategic mistake. The more shortcomings are made up, the more shortcomings are made up, and the shortcomings are made shorter and shorter. This has been done in the past few years, but the result is becoming more and more passive. The focus should not be on making up for shortcomings, but on strengthening strengths. With the new government leadership team in place, I believe that we will re-examine the ideas of semiconductor development, formulate new development strategies and measures, and invest more resources. There is a high probability that we will adopt different organizational forms from the past. This is my judgment. "
Prediction 3: EDA tools are not strong and sensitive chip development technology will become the focus of research and exploration. Integrated circuit technology will be more goal-oriented and problem-oriented, focusing on improving PPA of mature processes.
At present, the United States has used EDA tools, advanced semiconductor manufacturing equipment and other technical fields where it has absolute advantages as key "weapons" to restrict the development of China's advanced semiconductor technology. The new regulations introduced on October 7 have locked the development of advanced logic processes below 16/14nm, 3D NAND with 128 layers and above, and DRAM with 18nm and below.This also means that China will have to turn to chip R&D technologies that are not sensitive to advanced semiconductor processes and EDA tools at this stage, and focus on improving PPA with mature processes.
So, where can we start? In this regard, Professor Wei Shaojun pointed out that new devices, new materials, new processes, micro-nano system integration, and chip architecture innovation will be the three main innovation directions.
takes the three-dimensional hybrid bonding process as an example. It is to bond a logic wafer and a memory wafer face to face and then cut them. The resulting chip is formed by bonding the two chips together face to face, realizing the heterogeneous integration of the memory chip and the logic chip.
has also successfully helped the breakthrough of domestic 3D NAND with the help of this heterogeneous integration technology. For example, the long-standing X tacking technology in China is to independently process the peripheral circuits responsible for data I/O and memory unit operations on one wafer , while the memory unit is independently processed on another wafer. When the two wafers are completed, X tacking technology can bond the two to connect the circuit through millions of metal VIA (Vertical Interconnect Accesses) in just one processing step.
In contrast, foreign competitors have previously adopted the CNA architecture of placing peripheral circuits next to the storage array on the same wafer. In the past two years, we have shifted to a CUA architecture that first builds peripheral circuits and then grows the memory array on a single wafer.
Professor Wei Shaojun said: "Although it seems that the long-lasting architecture of two wafers bound together seems to be more expensive, and the CUA architecture seems to be cheaper, but when the NAND stack reaches more than 300 layers, you will find that the process of placing peripheral circuits underneath and then growing them becomes more difficult with more stacked layers, and the yield rate is very low. In the end, you still have to learn to live long-term. Major manufacturers like Samsung have also made it clear that if such a solution is used, it can achieve 10 00 floors. Why, this time the United States wants to put long-term storage on the entity list, because he sees that Chinese manufacturers are beginning to take the lead in technology. This shows that we are on the right track in terms of technology. "
Professor Wei Shaojun introduced "software-defined chip technology" as an example. In the past few years, the United States has implemented the electronic revitalization plan and promoted software to define hardware. It spent more than 70 million US dollars and mobilized more than a dozen units, several universities, and many large companies to do this, but the final results were not as good as China's.
"We did it earlier than the United States. We proposed this method more than ten years ago. If this method is implemented together with the hybrid bonding technology we just mentioned, the result will be a new near-memory computing technology. Storage and computing become very close without peripheral interfaces. This near-memory computing resource utilization is greater than 90%, data transfer power consumption is reduced by 70%, and memory access bandwidth is increased by 10 times. This will allow us to use mature processes to achieve higher than Mainstream processes are one to two orders of magnitude more efficient in computing energy. "Professor Wei Shaojun emphasized: "Computing energy efficiency is very important. If the computing energy efficiency of a supercomputer is not high, the power consumption will be huge. "
Therefore, at this stage, when the development of advanced technology is blocked, we can focus on chip technology that does not require the most advanced technology. Of course, we will still continue to pursue advanced technology, but the most advanced technology is not necessarily the best. The above examples also illustrate that our development cannot be stopped even if others want to. As long as we work hard and innovate, we can still move forward.
Prediction 4: Relying on China's huge market, a number of highly innovative products and solutions will be launched, promoting the construction of Chinese characteristic products and application standard systems, effectively easing foreign embargoes imposed on China and breaking containment.
According to WSTS data, global semiconductor sales will reach US$555.9 billion in 2021, and China remains the world's largest semiconductor market, with sales in 2021 of US$192.5 billion, accounting for 34.6%. Obviously, faced with such a huge semiconductor market, semiconductor manufacturers such as the United States are never willing to give up easily.This is also the advantage of China's huge market. So how can we make better use of this advantage?
Professor Wei Shaojun believes that we should "drive industrial development through application innovation." For example, in the 5G application field where we have advantages, the application of 5G has just begun, and eMBB is just an important application. What is really important is mMTC and uRLLC, which can be applied in agricultural modernization. We have 1.8 billion acres of cultivated land that cannot be touched, which is equivalent to 120 million hectares. Each hectare is 10,000 square meters. If you put a small sensor per square meter, you will need 1.2 trillion sensors. This was originally unimaginable, because the original wired method required digging trenches and burying cables to bury these sensors . Now with 5G, you only need to set up a 5G base station, and sensors of any square kilometer can be connected.
Obviously, China’s agricultural modernization alone will have a huge impact on semiconductors in the future. Now some places have begun deployment, not to mention the construction of many of our new smart cities. A large number of chips and sensors are in the process of deployment.
Professor Wei Shaojun said: "I think China needs to form its own standard system, which is more urgent than ever. The reason is that we have been learning since the reform and opening up of and , so we use a lot of things that have been circled by others. The product specifications have been defined, the product design method has been defined, and the product design path has also been defined. Well, we are just imitating others. In the face of such a huge country with a population of 1.4 billion, we don’t have a deep understanding of ourselves. Now that others have blocked us, we can actually innovate in a lot of ways. When China’s entire market of 1.4 billion people is mobilized and forms its own product specifications, product systems, product standards, and even design specifications, who should follow who? I don’t know yet. So we should not belittle ourselves and think that we are not good enough, not to mention that we are now entering intelligence. "
" The first industrial revolution solved the problem of mechanization. We extended our limbs. Later informatization solved the problem of the senses. We can see farther and the sound can be transmitted farther. The current technology revolution solves the problem of brain power. With the development of artificial intelligence technology, China is not much different from the United States. Therefore, the United States wants to restrict China at this time and not sell advanced chips to China, hoping to prevent the development of China's artificial intelligence. But can the United States really stop China's development? I don’t think so, we’ll see! ”
Prediction 5: Integrated circuit talent training will focus more on practical talents. The talent training program for outstanding engineers in the field of integrated circuits will be fully launched. However, how to realize the integration of production and education into will become the biggest challenge for the cultivation of integrated circuit talents in colleges and universities.
According to data, China's integrated circuit industry now has about 570,000 employees, and it will reach 790,000 in 2024, a difference of 220,000. Among them, the talent gap for chip design is now 120,000. One of the important reasons why there are still so many people missing is that the industry is developing too fast and talent training cannot keep up.
Professor Wei Shaojun also said helplessly, "The Ministry of Education appointed me to be the secretary-general of the subject evaluation group for the first-level discipline of integrated circuit technology and engineering. I am taking the lead in formulating a training system for large-scale talents. The more we formulate it, the more we find that our talent training cannot keep up. In the past few years, our talent training system has begun to move away from industry. An engineering major is far away from industry, so our next talent training will definitely focus more on applicable talents."
In addition, there is still a serious mismatch between the supply and demand of integrated circuit talents. Judging from the talent supply and demand index in major cities in the Yangtze River Delta from January to June 2022, the overall talent supply situation seems to be relatively optimistic. But a major flaw with these data is that they assume each job applicant submits only one resume. Obviously, this is inconsistent with the actual situation.
Generally speaking, job seekers from society will submit multiple resumes, and college students will submit even more resumes.If we assume that social job seekers submit an average of 2 resumes and college graduates submit an average of 5 resumes, it can be seen that the number of competitors for each position is less than 1. Why? This is the imbalance between supply and demand.
Judging from the requirements of integrated circuit companies for fresh talents, various ability requirements are continuously improving. Compared with the 2021 survey statistics, the degree to which fresh graduates meet the needs of enterprises in 2022 has also significantly improved: the proportion of satisfaction below 20% has dropped from nearly 22% last year to 9%; the proportion of fresh graduates with skill satisfaction of less than 40% has dropped by 10%, and the proportion of skill satisfaction between 60% and 80% has increased by 6%.
Wei Shaojun said that although the satisfaction of integrated circuit companies with talent capabilities is increasing, most of them are still not satisfied. Our colleges and universities often feel that the talents we train are very good, but in the eyes of enterprises, they are far from enough, and many of them are not what enterprises need.
What to do? Through the integration of industry and education. The State Council has issued many important documents before, such as "Several Opinions on Deepening the Integration of Industry and Education" in 2017. Later, an “Implementation Plan for the Pilot Project on the Integration of Industry and Education” was issued.
However, in Wei Shaojun’s view, the domestic integration of industry and education is not done well. Some efficiencies believe that the integration of industry and education is what vocational schools should do, and that this is not the job of universities. This is obviously wrong. "The country has now placed the integration of industry and education in a very important position, and colleges and universities have to do this work. There are two things here: practice and training. In the process of development, we are faced with how our enterprises and colleges can jointly cultivate talents. Our colleges and universities must re-understand the difficulties of the integration of industry and education, and re-find channels for the integration of industry and education."
"It is indeed difficult to make predictions, but I am still full of enthusiasm Confidence. China is in a period of industrial upgrading, and what we are doing is upgrading from the low-end to the high-end. What the United States is doing is industrial reconstruction. It is too difficult for a country with high income, high consumption, and low added value to produce low-cost, low-value-added products. For us, the future is to go all the way up, and we must understand the situation and strengthen our confidence in development.
Editor: Xinzhixun-Rurounijian This article is exclusively compiled by Xinzhixun
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-