Editor’s note: At the beginning of the new decade of the 21st century, Sohu Technology officially launched the “China Innovative Companies 100” series of lists and reports, focusing on 5G, AI, chip, manufacturing, retail, travel, social networking, corporate services and other fields In-house technological innovation and business model innovation, deep value mining for high-quality innovative companies and related industries.
“Technological innovation” has become a hot topic in the Internet industry. What kind of innovation is the innovation that a technology company should have? Through “China Innovation Company 100”, we try to answer these questions.
Recently, Sohu Technology held the first phase of the “China Innovation Company 100” chip series salon event. Liu Leibo, long-term professor of Tsinghua University and distinguished professor of Changjiang Scholars, Wu Youwen, vice president of Wingtech Technology, and Zhao Zhanxiang, partner and chief technology officer of Yunxiu Capital, discussed the new opportunities for domestic substitution under the global core shortage. This article is a wonderful view shared by Zhao Zhanxiang at the salon.
Zhao Zhanxiang, partner and chief technology officer of Yunxiu Capital, participated in the online salon discussion
Produced | Sohu Technology
Author | Liang Changjun
Edit | Yang Jin
Under the global core shortage, the domestic semiconductor primary market is still hot. Recently, at the first phase of the “China Innovative Company 100” chip series salon held by Sohu Technology, Zhao Zhanxiang, partner and chief technology officer of Yunxiu Capital, mentioned that in the past two years, investment institutions have begun to look at semiconductors, and many are rushing to pre- -IPO projects, but some institutions are not pure investment, but are highly speculative.
This wave of intensified core shortages does not seem to affect the market’s pursuit of semiconductors, but the impact on enterprises is real. Zhao Zhanxiang said that compared with the shortage of memory chips in 2016 and 2017, this time there is a shortage of cores in the entire industry, which is relatively rare, and has a different impact on each link in the industry chain.
For terminal manufacturers, relying too much on a certain supplier may be affected to a certain extent; design companies and Foundry (chip foundries) with production capacity or stocks on hand can adjust their product structure and prioritize production of high profitable ones. Products, earn more money; some start-up companies are expected to get the opportunity to enter the supply chain of major customers; the most influential companies are those that have no capacity before and want to take capacity temporarily.
For investment institutions, in addition to listing, mergers and acquisitions are also an important exit path, but the current domestic semiconductor mergers and acquisitions are not active. Zhao Zhanxiang believes that in a few years, the number of mergers and acquisitions by Chinese semiconductor companies will be very large, and the scale will gradually increase. Now there are three or four thousand domestic chip companies, and it is expected that in 10 years, there will be more than 50 companies with a market value of more than tens of billions. US dollars, in which many mergers and acquisitions will occur.
Different types of companies have different merger strategies. He said that mergers and acquisitions of digital chip companies are often for the deployment of new products or next-generation technologies-taking time to do it yourself, it is easy to miss the time window, and it is better to acquire start-up companies that have done better. M&As of analog chip companies are more to expand product lines. The more products and the more complete they are, the easier it is to be favored by customers, but a company often cannot do it all by itself.
In the salon exchange, Zhao Zhanxiang also shared his views on the core manufacturing of major Internet companies from the perspective of investment. He believes that domestic Internet companies must first solve the business model problem of making cores. Either turn the chip into a product that helps the company make money, or invest in an independent chip company outside and have a certain control over it. These companies can also use their own The platform is growing rapidly, and the latter is a better way.
Regarding the investment opportunities under the domestic substitution, Zhao Zhanxiang said that now semiconductor investment has entered the deep water zone, all the good fruits have been picked, and the rest are difficult to pick. He mentioned that since the beginning of this year, investment institutions are also focusing on areas with deep moats, low localization rates, and large capital needs, such as large chips such as CPUs and GPUs, and EDA.
In addition, IDM-based domestic memory chip manufacturers and third-generation compound semiconductor manufacturers have also raised large amounts of financing this year; automotive sensors, radars, and high-end chips in the industrial field are also very hot, with high thresholds. , Has the ability to expand horizontally in the future.
He also called on the state to have more policy support for chip companies that are willing to sit on the “cold bench”, difficult and long-cycle chip companies, and investment institutions should be more patient and pay attention to more early stage and greater value for China’s innovation. the company.
The following is Zhao Zhanxiang’s sharing record (deleted and sorted out)
Sohu Technology: You have been in contact with many startups. What impact does the current lack of cores have on them?
Zhao Zhanxiang:The impact of the lack of cores on each link of the industry chain, such as terminal manufacturers, intermediary agents, chip design companies, and Foundry (chip foundry), is different. For terminal manufacturers, relying too much on a certain supplier will be affected. For example, a certain chip of a mobile phone manufacturer has always come from one supplier. Suddenly the capacity of this supplier is tight. At this time, it is difficult to transfer to another supplier. Without a long-term cooperative relationship, the other party will not temporarily give you the capacity. .
When I made memory chips myself, it was normal to be out of stock, especially memory chips have a very strong periodicity. For example, in 2016, NAND FLASH (flash memory) was in short supply globally; in 2017, DRAM (memory) chips were in short supply globally, and the price increased seven times. If the agent has gone through this cycle, it can predict the situation of tight production capacity and stock up in advance.
For design companies and Foundry, if it is a company that has production capacity or stocks on hand, it actually hopes that it lacks cores and raises prices, so that its profits will be higher. If there is a relatively stable supply of production capacity, the response to the shortage of cores will be adequate. Although the production capacity cannot be further increased, the proportion of some high-margin products can be expanded. In this way, although the overall sales revenue volume does not grow so fast, the gross profit increases, and the final profit margin will increase.
In addition, under the current situation, large customers are also willing to introduce new suppliers, which is just an opportunity for some start-up companies. But it does have an impact on small, low-capacity companies, and these companies still have to reserve cash for the winter at this point in time. The biggest impact of the lack of core is that companies that did not have production capacity before, and want to temporarily take production capacity, now it is difficult to squeeze production capacity.
Sohu Technology: There were also global core shortages in 2016 and 2017. What is the difference between this time and the previous ones?
Zhao Zhanxiang:The semiconductor industry has been cyclical since its inception, and it is difficult to fundamentally solve the lack of core tide. The core reason is that it takes more than a year for semiconductor production capacity expansion, but demand is increasing rapidly. In 2016, NAND FLASH shifted from 2D technology to 3D technology. 2D production capacity was partially withdrawn and turned to 3D. However, the yield rate of 3D did not increase, resulting in a decrease in market supply, and increased demand for mobile phones and data centers. PC SSD (solid state) Hard drives) have increased in capacity and demand soared. This is the same as the memory shortage in 2017, which is an imbalance between supply and demand.
This time there is a shortage of cores in the entire industry, which has a lot to do with the epidemic and the rapid spread of 5G mobile phones. Market demand has increased and supply cannot keep up. This kind of large-scale core shortage is relatively rare, and companies need to prepare in advance to stock up in advance when there is a sign of core shortage. In addition, the financial strength must be large, and a long-term cooperative relationship with the fab is also required.
Sohu Technology: How do you view the opportunities for domestic substitution from the current shortage of cores? How to treat the current level of localization?
Zhao Zhanxiang:What domestically produced substitutes have to solve is the stuck neck problem. Now the good fruits of the semiconductor industry have been picked, and the rest are more difficult to pick, such as CPU, GPU, DSP, FPGA, and memory chips NAND FLASH, DRAM and other fields with a low localization rate. These localization rates in the next ten years need to exceed 50% or even higher, which is a very difficult task.
Semiconductor manufacturing equipment and materials currently have the greatest impact on the neck. China can now build lithography machines, but it is relatively low-end and can only do 90nm. Once the key materials, key EDA tools used for designing chips, and IP are blocked, it will have an impact on the industry. It is also very large, and it is urgent to realize domestic substitution.
On the other hand, my country’s innovation fields need to be synchronized with the world, such as quantum chips, photonic computing, and new sensors in mobile phones. Now we are faster in the iterative speed of new technologies than abroad, the pace of work is faster, capital investment, human investment are stronger, there are reasons to be more confident in our level of innovation.
Sohu Technology: Semiconductor investment has been very hot since last year. Under the tide of domestic substitution, what are the relatively large investment opportunities?
Zhao Zhanxiang:Yunxiu Capital has been paying attention to the semiconductor industry very early. According to our observations, there were not many investment institutions that looked at semiconductors before 2017. After the ZTE Huawei incident, many institutions began to look at this field. Semiconductors are not only a policy-oriented market, but there is really strong market demand. Large domestic customers used to have only the top three suppliers in the world, but now they also want to supply the domestic supply chain. This is a big market opportunity.
After the establishment of the Sci-tech Innovation Board, semiconductor companies performed very well on the Sci-tech Innovation Board. Although the number of the SCI-Tech Innovation Board accounted for only 15%, the total market value exceeded 30%. Last year, the market value increased by an average of 40%. Due to the popularity of the science and technology innovation board, the amount of investment in the primary market of semiconductors last year exceeded 140 billion yuan, a nearly five-fold increase compared to 2019. Last year’s primary market equity investment enthusiasm was mainly concentrated in Pre-IPO projects, but their valuations are very high, and the return on exit may not be that large. Everyone is going to grab the Pre IPO quota. It is no longer pure investment, but very speculative.
This year, many investment institutions are focusing on areas with deep moats, low localization rates and relatively large capital needs, such as large chips such as GPUs and CPUs, and companies in areas such as EDA software. IDM’s financing this year is also very large. Many domestic memory chip manufacturers and third-generation compound semiconductor manufacturers are all in the IDM model. Subdivision areas have also attracted much attention, such as automotive sensors, radars, and high-end chips in the industrial field. The threshold is very high, and the future has the ability to expand horizontally.
Sohu Technology: For investment institutions, mergers and acquisitions are also an important exit path. What do you think of the M&A trends in China’s semiconductor industry?
Zhao Zhanxiang:Last year, global semiconductor mergers and acquisitions were very hot, and the overall capital estimate exceeded 100 billion U.S. dollars. M&As of overseas giants often lie in their strategic layout. There are not so many domestic mergers and acquisitions, but now there are three or four thousand domestic chip companies. In 10 years, there will be more than 50 companies that will become tens of billions of dollars. There will be many mergers and acquisitions. Integration occurs mainly due to three factors:
First, the market value of leading companies in the industry is relatively high, and they need to install new assets to support their stock prices; second, digital chip companies need to lay out the upstream and downstream industrial chain, and analog chip companies need to increase product lines in order to continue to expand; third , The future liquidity of small market capitalization companies on the Science and Technology Innovation Board may not be so good, and capital withdrawal after listing may not be ideal. It is better to sell to large companies for cash in advance. Therefore, the number of mergers and acquisitions of Chinese semiconductor companies in the future will be very large, and the scale will gradually increase.
Sohu Technology: If domestic chip companies become bigger and stronger, are mergers and acquisitions necessary? What are the differences in M&A strategies for different types of companies?
Zhao Zhanxiang:The threshold of the semiconductor industry is very high. If you do not acquire, you need to do it yourself from scratch, and the time and cost to spend is very high. So now many companies are willing to go overseas mergers and acquisitions, because overseas companies have already accumulated a lot. However, the ultimate goal of the acquisition is either for technology or for the team, but if it cannot be digested, the value of the acquisition will not be that great.
From the perspective of the development of technology companies, it is common for large companies to acquire small companies, but the acquisition strategies of digital chip companies and analog chip companies are different. Digital chip company acquisitions are often for the layout of the next product or next-generation technology. There are many new directions, and it is impossible to invest heavily in research and development for each. Instead of spending four or five years trial and error, it is better to acquire a ready-made company so as not to miss the time window.
Many times analog chip companies merge to expand their product lines. China’s largest analog chip company now has thousands of product lines, and giant companies such as TI and ADI have tens of thousands of product specifications, which is convenient for customer supply chain management and has a great advantage in winning customers. Therefore, for domestic analog chip companies, the future will definitely increase product lines through mergers and acquisitions.
Sohu Technology: From an investment point of view, how do you think about the Internet to make cores? What problems will you face?
Zhao Zhanxiang:Internet companies must first solve the business model problem for core manufacturing. Internet companies may only need hundreds of thousands or millions of chips a year, while the chip industry has technological iterations every year, and the annual investment in new products is high or even doubled. How can Internet manufacturers bear the cost? Strength to feed the chip team? This is difficult commercially.
Foreign Google and Amazon are very successful in making chips. They are not made to digest the chips by themselves and reduce costs, but are strategic layouts. Through self-developed chips, the number of virtual machines supported by a single machine increases, and then the services are sold to cloud computing customers. , To recover the cost of the chip through service.
How do domestic Internet companies recover the cost of making chips through services? Either through business model innovation to turn the chip into a product that helps the company make money, or invest in an external independent chip company to have a certain degree of control over it, and let them grow quickly with their own platform. This is a better way.
Sohu Technology: Do you have any suggestions for promoting the development of China’s chip industry and domestic substitution?
Zhao Zhanxiang:The Sci-tech Innovation Board needs to continue to encourage high-quality loss-making companies to be listed. The development cycle of chips is very long, and it is difficult to generate tens of millions or hundreds of millions of profits in the short term. If these companies are required to make profits before they can go public, everyone is not willing to make difficult chips, but will be short and fast, attracting speculative people. , This is bad for the industry.
The country needs to have some policy inclination for companies that are willing to sit on the “cold bench”, with long cycles and high difficulties. Many institutions now tend to invest in Pre IPO projects for short-term IPOs and exits. This is not very helpful for many innovative companies or companies that really need money. Therefore, I also hope that the exit cycle of investment institutions can be longer and be patient to invest in earlier and earlier, Companies with greater innovation value to China.
You must log in to post a comment.