During ICCAD, EE Times China interviewed senior executives from EDA, IP, design services and foundry companies and discussed the development and prospects of China's semiconductor industry and possible obstacles in its development.
At 2017 ICCAD Summit, Professor Wei Shaojun, chairman of IC Design Branch of China Semiconductor Industry Association, gave an authoritative summary of 2017 China's semiconductor IC design industry.
Statistics show that as of November 2017, there were a total of 1,380 IC design enterprises in China, an increase of 18 over 1,362 in 2016, with a small overall change rate.
Professor Wei Shaojun believes that the large increase in the number of design enterprises in China by more than 600 in 2016 is both inevitable and contingent. The inevitability is reflected in the huge driving effect brought by the introduction of the "Outline". Incidentality is reflected in the unsustainability of this substantial increase in number. In 2017, the number of enterprises is basically flat, reflecting that the development of this industry has to some extent reverted to rationality and with the passage of time, a reasonable reduction in the number of businesses in the future is also expected.
From the sales point of view, according to IC Design Branch statistics, the industry-wide sales in 2017 is estimated to be 194.6 billion yuan, up 28.15% compared with 151.8 billion yuan in 2016. China's IC design industry in recent years is the fastest growing: in 2016, the growth rate was 23.04%. In accordance with the exchange rate between the dollar and the renminbi at 1: 6.65, the annual sales reached 29.263 billion U.S. dollars, increasing China's proportion of the global IC design industry to 30%.
In terms of IC product mix, the number of companies engaged in communications, multimedia, navigation, power and consumer electronics is up from 2016, while the number of manufacturers in smart card, computer and analogue is decreasing. Among them, the outstanding performance of communication chip design companies increased from 241 in 2016 to 266, corresponding to the increase in total sales by 31% to 900 billion yuan. Consumer IC products followed in second.
Jing-Yang Zhang, CEO of Moore Elite, was optimistic about some small smart hardware and wearable devices, that these points will bring sub-chip needs. "Before the chip is the market first, and then do the chip service market, but now that we see the new trend is the terminal first, many system vendors will kill the chip areas, such as Xinjiang, from the earliest to do drone is now pursuing the autonomous control of its own core chips and the security and stability of the supply chain," he said. "At the same time, the patterns of some chip companies are also changing. The core of the entire company is the chip, but the system behind it is directly facing the consumer. For the terminal market, this application may not be particularly large shipments, but it will bring a lot of new ideas and direction for the semiconductor industry."
He Ying, director of core technology products, shared the viewpoint from an IP provider's perspective. He said was optimistic about the data encryption industry. "There are also technologies behind digital encryption worth noting, such as blockchain; at the same time, we also concerned about the national security chip encryption field, with domestic IP made for domestic chips," he said.
The data and results are exciting!
During ICCAD, I interviewed senior executives from EDA, IP, design services and foundry companies and conducted dialogues on the development and prospects of China's semiconductor industry and possible obstacles in its development.
**1. Macroeconomic Prospects and Market Hotspot in China 's Semiconductor Industry ** AI, Automotive, IoT, Security and 5G are the consensus of most guests.
Lin Jung-kin, Synopsys' vice president of global operations and Asia Pacific president, first expressed his opinion: "I think there are two attractive topics in China from 2017 to 2018, one is the wide range of AI applications and the other is the automotive industry. There are many applications in two years and these two types of opportunities are very promising. "
Dai Weimin, chairman and president of VeriSilicon, gave a more detailed explanation of the AI applications and the automobile industry. He believes that in the field of AI, a AI application in the field of monitoring, that is, security, is needed. In the car, electric cars will be China's strengths, a number of brand-name car prices have begun to intervene, and introduced the Chinese standard. "Overseas car companies also use Chinese standards, and this is fantastic, and we have mastered the standards. Some AIs in China are getting involved in the AI algorithm, and the question is how to become a chip," he said.
TSMC is the largest semiconductor foundry company in the world and Luo Town Ball, the company's vice president of business development, said: "I think China is really in a very good momentum at a macro level. Now that IoT, automotive electronics and security are growing rapidly both in China and the rest of the world, China is simply saying that in the past few times China did not catch up with its capabilities and financial resources. In the future, in these three applications, China and other countries, The area starts at the same time and even takes up some advantages as we have a vast market. Automotive electronics will be China's largest market, IoT applications and population base. China in the field of these applications of ICs, three to five years later, certainly have different results.
"In the field of communications, China has begun to take the lead role in setting the 5G standard and combining communications with IoT to be a big market. Now, there are a lot of calculations to be done at the terminal, and the entire semiconductor, memory, processor There is a great demand, coupled with the security requirements, the overall market is very large.This is not just to say the general IoT, car networking industry is a part of IoT.China will be the leading position. "This is Kilopass sales vice president President Lee Chu's point of view.
Ms. Xu Yun, general manager of Cadence China, said that Cadence is optimistic about the development of the entire IC industry in China. In particular, the next round of major funds has also passed the examination and approval, and the investment in this industry will increase. "In terms of top apps, several key applications we'll see in the coming years include: The first is the AI direction and the second is the Internet of Things, which includes the Industrial Internet of Things. The Industrial Internet of Things may be better application space; in addition, there are automotive electronics, smart car concept is very popular in the country, we also see a lot of realistic opportunities. "It is reported, Cadence has been with a number of advanced smart car manufacturers to develop a wide range of cooperation.
*Part of the guests. * **2: How does China's domestic IC achieve or catch up with the international advanced level? ** Professor Wei Shaojun made clear in the keynote speech of ICCAD that to achieve the 2020 growth target of "just to be established", it will be difficult to achieve by the current products alone. To achieve the goal, one is to make major breakthroughs in bulk products, such as CPU / DSP, memory, and FPGA products. The other is to increase product innovation.
Lin Rongjian to market size and application scenarios, for example, that China exists in the field of mobile communications, surveillance and other fields. At the same time, he is very optimistic about the corresponding application of AI, divided into two categories: one is the AI chip security IP, the other is the application of AI, that these two points complement each other in 2018 and even in the next few years there will be a great increase. In the past few years, China has mastered the second half of the mobile field, but now China has more applications that can drive more technological developments.
"There are more than 1,300 IC design companies in China, most of them still maintaining small and medium-sized, distinctive conditions. Are these different companies able to focus on one point rather than doing homogenization? In Taiwan's experience, many manufacturers do something from the beginning is similar, but slowly classified, but also some communication, we grasp a hot spot to go forward, do not cross each other.China now have a problem hot As soon as we get together, one can get the capital to try another, but this process does not guarantee that you now have a truly competitive company because the technology needs to be accumulated and studied, Once there is a problem, it is difficult for the enterprise to expand. "The enterprise should have its own mind and I am worried that price war will be the first to proceed, rather than the gradual accumulation of technology, and that there will be no good chance in history." Lin Rongjian gave his suggestion.
Dai Weimin answered this question from the industrial level, thinking that there are two dividends: one is capital dividend, with funds; the other is to make dividends. From his point of view, the dividends that Chinese IC design companies will see in the coming year will be great. At the same time, he urged the fund to dare to invest in high-end and long-term companies.
Luo Town ball to TSMC contact customers made a reply. "In China, there are always new companies coming into contact with us and the products we make are all kinds of things, and this is what I am talking about passionate and you will chase for a long-term and idealistic goal. Now I see more powerful, There are passionate, funded, and the development of consumer products are also earlier.Consumer covers a wide range of consumer safety, which is a big hot, AI is also on the one hand, "he said," 360 degrees 8K I see a lot of admirers for the integration of image quality in a system, the gap in light can be compensated, and TSMC has a very wide coverage in the world and we probably know of new products. China has accumulated more than a decade of technology At the time of launch, the industrialization was now poor, and the knowledge and experience that domestic enterprises must possess in industrialization and industrialization are not enough. This may be the next step to be strengthened. "
Lee Ch believes that industrialization must work together and that in China we are still playing a role of competition, not enough cooperation platforms, or not enough resources for exchange or cohesion, and resources can not be combined. "I see a great market for microprocessors and sensors and see how they fit together, and this is also a new area for our guests, who come in contact with each other, memory and sensors together.This market is very large, more and more products in the future. "
Xu Yun believes that China has seen resistance in overseas mergers and acquisitions, local IC design companies are training internal forces, down to earth to do a better job of the main industry. "First of all, there is one piece of data for reference. In the field of AI, China is the second largest AI economy in the world with three aspects: one is investment, the other is the number of companies, the third is the number of patents owned and the third is ahead very much. Using HiSilicon as an example, unveiled the Kirin 970 chip in the market by a lot of recognition, the internal use of Cadence's DSP. "She said," In addition, the country has been advocating autonomous controllable direction, in this There are also many opportunities for local Chinese enterprises in the field, such as AI, image recognition, and security. China has many advantages in the system direction. For example, Hikvision and UOB are global systems in the field of intelligent monitoring How can we take advantage of our system advantages to develop our chip area, I think it is our future direction. "
He Ying refers to the shared cycling that is one of the four new inventions in China. He believes that the current development of cycling in China is better than any other country in the world. However, chips used today are still foreign solutions. Chinese local companies began to design miniaturized smart chip, as long as the market, there are applications, in this regard to catch up with foreign advanced companies and technologies.
**3: How does an EDA company fit in the AI era ? ** At present, the major EDA companies are expanding its coverage of services, in addition to EDA tools / IP and other traditional business, but also extended to other services. At the same time, EDA tools in the artificial intelligence era what will happen under the trend of development, the tool will become more intelligent?
I remember very clearly that at an IC conference in Tianjin about two years ago, many of my familiar friends asked me if your company was not working properly and how to make software from EDA and IP. After more exchanges, more meetings will be held. Is a positive look at this issue, "Synopsys Lim Jung-kin replied," I think this is right. Seems to extend the scope, but we think the same is the same, with the needs of the industry go.IoT development involves security, when you talk Design, let alone security is a big chunk of our expansion is to follow the logic of business needs to go forward, we have a relatively simple business balance, where more demand for more investment.
"Our perception of security is this: If it's too late for security to wait until it's been applied to the system, because there's no way to completely block the outside world after the system goes live, you have to think about it from the ground up, We have to put the system architecture into. Our solution covers the beginning of the entire software, there is a corresponding security program, so that no matter what the scene, there is basically a certain degree of security. "He explained Synopsys' Security Software Business.
"Synopsys has seen a lot of growth in China in recent years, and we know very well that we can not do anything, especially when there are some things that our partners do well. Many of our customers need design services or packages. We work with VeriSilicon, or we With the TSMC in the early stage of a lot of cooperation in research and development.We do the most good part, the other should basically cooperate fully.
Lin Rongjian that AI is the process of intelligent, corresponding to different applications, EDA is one of the applications, so EDA more intelligent. "We've invested a lot in AI, and even AI in different product divisions in different geographies." Our IT department has developed a common platform called 'Common Machine Learning Platform' that is designed to get different products Development process does not need to do repeated design.EDA in the process of running different software itself has huge amounts of data, the use of such a platform quickly back to the machine learning, which helped our development.Many important users with us Designing for such a solution, after a few months of machine learning, now takes only 1-2 weeks. "
Another area is the design of AI chips. "AI chips will certainly use a heterogeneous architecture. A lot of computing, coupled with enabling heterogeneous design to run, usually with specialized tools. The second is verification. Previously verified chip verification is now AI will use On a variety of different products, including automobiles and autopilots, we have to be able to ensure that life-threatening in autonomous driving scenarios does not occur, so verification is more challenging for new AI applications, "he said.
Cadence Xu Yun that basically several mainstream EDA companies are expanding service areas, is to comply with the needs of the industry and customer needs and changes. "To catch up with Moore's Law, the accumulation of many design companies did not do all the IP as before, can be completed through their own design process.Now the chip complexity is getting higher and higher, we must solve how to cooperate with the development of Moore's Law In this premise, several EDA companies are expanding to more areas.One is IP, we also see each of the major customers, basically need the cooperation of IP.Another is to verify that now The synergy of software and hardware, including the growing need and complexity of software development, is where all of us are headed, "she analyzed." In addition, the system is becoming more complex and has raised the bar in every other direction, Such as packaging, PCB system-wide design, will bring more challenges and new directions for Cadence, in addition to cooperate with the development of Moore's Law, the second major direction is that the system is driving the entire chip design Development, we have to consider the system-wide optimization of software and hardware, the overall integration of all directions. The accumulation of experience in the application of the system and design services for system applications are the general directions for the future development of Cadence. Combined with Cadence's growth strategy, the company recently set up a stand-alone company in Nanjing focusing on IP and systems design services.
In response to the needs of the AI era, Cadence offers solutions in both CPU and DSP, including DSP IP with Neural Network for AI. In the AI chip verification, Cadence offers multi-engine program to ensure the final verification can be completed effectively, but also hardware acceleration simulation equipment. "How does EDA itself evolve in the AI era? Whether it's implementation or verification, or traditional analog and mixed-signal circuits, we all have algorithms that use deep learning and machine learning to optimize our software, and more importantly, if everyone's design data Being able to build a public big data foundation is a great opportunity for us to make a new breakthrough at EDA, turning it into a learnable model that allows new companies or new design projects to draw on the experience of past projects and You can learn from other friends. One question is how to share data, because these data for each design company is their own intellectual property, at this point, we also want to maximize their ability to cooperate and allow customers Under the premise of providing help. "
*More guests. * **4. The Roadmap for the Next Development of China Foundries ** In recent years, both overseas foundries and local foundries have been building factories in China quite frequently and are surging.
"The same is true in Taiwan 20 years ago when the most exaggerated issue was the hoarding of a pig, which was actually a craze in Taiwan. In China, because of the NDRC and the government, the whole situation It will be a lot better. You want to build a factory is not difficult, what technology to do, where your customers, this is difficult, "TSMC Luo ball bluntly," Our factory in Nanjing, not yet completed before the completion of all road map design, Electricity how to cloth, how to get gas, how to drain the water, what technology have been very clear.However, found that there are some specialties in the country.Would you like to plant how much? He said 100,000 12-inch like; what is What is the product process? Anyway, that is 100,000. Finally, I want to know where the customer. A domestic industry should be solid, so strong, sure to put money, manpower and material resources to figure out what to do, do What technology, what to do to distinguish between the next step to do, the product to whom? This is my advice to the domestic industry.
In view of the future development of TSMC in China, Luo made an analogy with "the army is not moving, the grain is advanced first". Forage is the order, why TSMC plant in Nanjing? It is because China's IC design has come to this step. TSMC in China now uses 16nm, 12nm projects more than 20 companies need to serve these customers base, but also to provide local services the most efficient time. "Naturally, this will be the next step, and our next step is to see exactly what kind of products we can offer to Chinese design houses, and whether Chinese design firms can make very competitive products with it. There is demand. "
Founded in 2010, Shanghai Huali Microelectronics Co., Ltd. (HLMC) intervened in the 12-inch foundry business in China earlier. Its vice president Shu Qi shared the history of Huali Construction: one phase of HLMC has been successfully completed, and 35,000 units of production capacity Even to the first half of next year are full, this demand does exist. When we first built the factory, China's design firm was weak at the time and has now reached the second place. China's future development is indeed good for us, so our two is 40,000, the main concern at 28nm. Three, four are also prepared, while the three piling has ended, but also 40,000.
"I think this is a good thing for flourishing cities such as Nanjing, and if competition is getting stronger and stronger, we can create a better platform for domestic design enterprises and have more choices. Challenge us Is to choose more, we must make the process well, "Schuch said," When I visit the customer, the customer usually have a proactive and passive requirements.Activeness is the customer's new products come out to require such a technology To achieve, for example, he asked for power, smaller size, faster; passive requirement is that this node has done a good job, but why go forward? The cost of each node is very high, but if not forward Go, competitors will move forward Passive forward is also a driving force, if there is no 14,16 nm, many people will not go forward, the entire development may be limited to the 28 nm limit, there is no impetus for development.
**5. New Requirements for Design Services **
With the development of IC design industry in China, new requirements have also been put forward on design services. Dai Weimin from VeriSilicon attended the interview as a representative of Design Services.
"First of all, we absolutely do not compete with customers, secretly products. We have a perfect IP core, and in some areas than the United States to do well," Dai Weimin stated that "the company is now providing cloud services more and more strong The power is bigger and bigger.That is a change.We used to serve a lot of chip companies, now the objects that need to be served have become Ali, Tencent, Baidu, Amazon, Google and other types of companies, they want to do a high voice chip. They are leading the industry and need customization, and now the cloud is full of video and the entire industry is dominating. Some customers are now moving toward the system level, which is in line with this trend. "
Dai Weimin concluded: "Chip companies go up, first a board made chips, write software, do a lot of system things.When the chip companies go up, the system company down, but do not have the chip Team, so look for our design services company. Up, down are likely to find us. "
**6: IP Manufacturers' Opportunities in China ** IP needs related to all aspects, here to memory and hybrid IP as an example.
Kilopass currently focuses on memory IP in China, and the company, Lee Chu, sees the demand for memory chips evolved from simple storage to security in recent years. "It used to be a simple key, and nowadays it has its own system for storing keys, and memory is an integral part of the processor." China has been more concerned with flash in recent years. "He said that" "In China, this part has not been studied, mainly requires a lot of long-term accumulation of technology, which is relatively weak domestic. Do IP less lovable, unlike R & D products, you can sell millions, tens of millions, IP is Basic work needs to develop slowly and hopefully in the future to make it more in-depth. "
Core motion technology to provide hybrid IP, the company Ying Ying that the IP demand within the next one to two years, including the following hot spots: On the one hand IoT chip for some new technology nodes, requiring lower power consumption, a new IP demand. This demand reflects the driving force behind, we have to rush in front of the market to respond to the IP ready; the second is high-performance applications. Now we know that regardless of AI or other areas of HPC, a relatively large bottleneck is the bandwidth, the current approach is to pursue the direction of DDR, but in the new application of this speed has been unable to meet the requirements, and need A faster interface, now only used in graphics cards, which is why some companies are leading the AI in most other companies, because they have the key technology, which is where we want to work; and finally, we IP cost advantage, and our service.
**7: Differentiation of China 's Semiconductor Industry ** Differentiation will not be transformed into the core competitiveness of all enterprises?
"Mentioned the difference, how to use Internet companies to increase the viscosity of the customer? There are many is to be built on the hardware system.Many system design companies put forward new requirements for hardware, but also because of the application requirements of these changes caused by The more and more customized needs of the hardware, which is the interaction of each other. This fact has a very big impact on our industry, which is why we put forward the 'system design to achieve' big strategy for us, the whole company The thinking of future development is indeed affected, we are also making the appropriate adjustments and changes. "
"In the automotive field, for example, more than just basic work, we are more experienced in application. We cooperate internationally with state-of-the-art smart car companies and we have also acquired a number of Automotive electronics or smart car applications and we are now working in this direction, despite some bottlenecks, but also hopefully for the entire automotive electronics industry to do some system-level support, "Cadence Xu Yun said.
Autonomous control can also be understood as a way to achieve differentiation, in this regard, Shu Qi shared his point of view. "Self-control is the core technology you have to master some of the early start-up small and medium companies, not so much energy in the early days, there is not so much independent intellectual property rights, he had to rely on some powerful collaborators to help him, This is a lot of technology and all aspects of the hands of others, after the completion of this product to continue to develop more difficult, but this is the process, from the beginning are so go.China's development will one day be autonomous Controllable, for example, Haisi is not completely self-control at the beginning, there are many technologies that rely on a number of manufacturers, but after a period of time running out of his technical strength accumulated, the business is also done, research and development into more and more, to Finally, we can make his own alternative chip, he mastered this technology, including us, too, we must master the core technology, we must develop our own independent and controllable brand, in order to support the vigorous development of IC design companies.
Zhang Jing Yang also shared his point of view: To go beyond the previous company, go it the same way to catch up may never reach, replace Qualcomm is definitely not the same way to replace, we must make good use of our home advantage. Since almost all of the world's electronic products are made in China, we must find differentiated opportunities here. Phone has been very smooth industrial chain, we use the most convenient, but in reality there is room for optimization of each application, you need to domestic manufacturers is not simply to do to replace foreign products. The beginning of this road is very easy, but more and more difficult, come back will encounter a patent, encountered other bargain. It may be difficult to start with innovations and define new products yourself, but it is often difficult to do so. A difficult task can start to be started and some progress can be made. If you continue, you will have barriers behind , Then foreign companies want to come to us too difficult.
**8: The Most Necessary Problem to be Solved in China 's Semiconductor Industry for Sustained Development ** The following are the key elements that the honored guests believe will most need to be solved or are currently the most deficient for ensuring the sustainable development of the Chinese semiconductor industry and their proposals for action in 2018 and beyond.
Dai Weimin: I think it is a talent bonus, there must be a special fund to start a startup.
Lin Jung-jian: It is not easy for technology to break through now. Technology is what people make. People are a big constraint and this must be solved. Need to plan from the entire education system, from a lower level to start.
Luo ball: I think there is Chinese government support, there are funds, a lot of people, very excited, the resources have to see. Everyone must have the entrepreneurial spirit, do one thing should focus, do a good job, strong and true.
Lee Chu: Beyond passion, be patient and keep going. See how it works with universities. Hopefully, the industry will continue to support university research and development. Many cutting-edge technology or research and development from colleges and universities, I hope more attention.
Xu Yun: From the current point of view, China's semiconductor development trend has been unstoppable. The most radical driving force for China's potential development is the huge market and the next largest development of data, which is different from the United States and any other country in the world. the value of. We should invest more in AI or big data to really transform our data into resources that are useful to us. In this regard, we should increase our chances of investing more in Europe and the United States.
He Ying: The problem that needs to be solved is how we become game makers. We need more participation in the formulation of international standards and rules in order to have a say in the world chip stage.
Shu Qi: 2018 is indeed a very good opportunity, both from the market and from the opportunities exist, so all the chip companies, including our company, should do their own thing well, not only to invest in research and development, and should be steadfast Do it. I think the media should make more proposals with the state. At the national level, we should see that this opportunity exists. We should give more support to some policies, especially some start-ups. Countries should give their support to start-ups, especially newly established and well-thought-out enterprises, and do not be afraid of failure because in many cases success is after failure.