Roundtable Discussion: The Second Half of Domestic Substitution: Paths and Challenges from Single-Point Breakthrough to Full-Value Chain Penetration | 36Kr Future Industries Summit 2025

On September 10th, the 2025 36Kr Industry Future Conference, hosted by 36Kr, grandly kicked off in Xiamen, China. This conference has joined hands with the "China International Fair for Investment and Trade" hosted by the Ministry of Commerce. With the core theme of "In the Era of In - depth Cultivation, the Tides Surge in the Land of Abundance", it has made every effort to create a high - standard, high - value, and high - influence industry event that combines national vision, industry depth, and market popularity.

The conference closely aligns with national strategic guidelines and the forefront of industry development. It focuses on five core sectors: artificial intelligence, low - altitude economy, advanced manufacturing, new energy, and mass consumption. It brings together top industry players to discuss development paths and outline the future of the industry. During the two - day agenda, with the "industry cooperation chain" as the logical main line, the conference focuses on the collaborative mechanism among the government, capital, and industry. It delves into how to break down barriers, integrate resources, and precisely address the pain points, bottlenecks, and constraints in industry development.

36Kr organized a round - table discussion titled "The Second Half of Domestic Substitution: Paths and Challenges from Point - by - Point Breakthroughs to Full - Chain Penetration" at the Industry Future Conference. Zhao Cong, a partner at Longding Investment, Zhang Gaonan, a managing partner at Huaying Capital, Yang Yang, a managing partner at Shunchuang Industrial Investment, Tang Zujia, a managing partner at Hechuang Capital, and Liang Feng, a managing partner at Shengyu Investment, participated in the discussion.

The following is the content of the round - table discussion, compiled and edited by 36Kr:

Zhao Cong: First of all, I would like to thank the organizer for the invitation. I am also honored to have this discussion with four very experienced guests. Let me introduce myself briefly. I'm Zhao Cong from Longding Investment. Longding Investment is an investment institution that has long been deeply involved in the entire semiconductor industry chain and has gradually extended to the fields of new energy, automotive electronics, and industrial software. Our core approach is to focus on the industry and pay great attention to industry synergy. Therefore, most of our investment targets and invested projects can form in - depth synergy. Additionally, it has recently been announced that our company is acquiring a listed company to further enhance the institution's ability to drive both industry and finance. Next, could you all please introduce yourselves and your relevant institutions?

Zhang Gaonan: Hello, everyone. I'm Zhang Gaonan from Huaying Capital. Huaying is an established VC institution since its establishment in 2008. We have gradually upgraded from digital content to consumption and then to technology. Currently, we are a relatively comprehensive fund. Our main directions include technology and consumption upgrading. Our past management scale was approximately 12 billion yuan. I am mainly engaged in technology - related investments.

Yang Yang: Hello, everyone. I'm Yang Yang from Shunchuang Industrial Investment. Shunchuang Industrial Investment is a wholly - owned state - owned enterprise in Shunyi District, Beijing. It was established in 2023. Our main investment direction is to support the five leading industries in Shunyi District, namely new - energy intelligent vehicles, aerospace, medical and health, third - generation semiconductors, and intelligent equipment. Our investment targets are mainly concentrated in the fields of high - end intelligent equipment and high - end manufacturing. At the same time, there are certain requirements for projects to be implemented in Shunyi District. If there are relevant cooperation opportunities, I hope you can make more recommendations. Thank you.

Tang Zujia: Hello, everyone. I'm Tang Zujia from Hechuang Capital. Let me briefly introduce three labels for Hechuang Capital: 1. We focus on early - stage hard - technology investments. Since 2012, we have been concentrating on early - stage hard - technology investments, with our investment directions focusing on electronic information and biological information. With the technological evolution from informatization and digitization to intelligence, humans are creating a digital intelligent world that maps and interacts with the real physical world. Our investments are mainly centered around this theme. 2. We maintain a stable level of activity. Hechuang's annual investment frequency is relatively stable and is not significantly affected by the financial environment. We basically invest in a dozen to twenty - odd projects each year. In 2024, we invested in more than twenty projects, and so far this year, we have invested in over ten projects. 3. We pay more attention to the impact of cycles on investments, including the economic cycle, financial cycle, industry cycle, technology maturity cycle, and the growth cycle of the enterprise itself. Through studying the relationships between these cycles, we hope to help our investors improve capital efficiency as much as possible.

Liang Feng: Hello, everyone. I'm Liang Feng from Shengyu Investment. Shengyu Investment is an investment institution with a profound industry background. Since its establishment more than twenty years ago, it has continuously focused on hard - technology investments, mainly in the four sectors of semiconductors, robotics, commercial aerospace, and innovative medical devices. We have always adhered to the investment concept of "in - depth vertical" approach. The so - called "in - depth" means to be deeply involved in the industry, to serve and accompany excellent startups with the mindset and positioning of an industry partner, to incubate technological innovation through VC, and to promote industry integration through mergers and acquisitions. The so - called "vertical" means that we do not select and judge projects from a single point, but rather sort out high - value nodes and high - growth enterprises from the perspective of the industry chain. At the same time, we pay more attention to thinking about the ability connection, resource integration, and chemical reactions between enterprises around the leading enterprises in the industry chain to support the iteration and development of the entire industry chain. Thank you all.

Zhao Cong: Thank you all for your introductions. Now, let's get back to the topic. Today's theme is the second half of domestic substitution. First, I would like to ask each of you to help us review the achievements of the first half of domestic substitution. Each of you can give an example or a specific data about the progress of domestic substitution in the fields you have tracked or are concerned about in the past one or two years. Let's start with Mr. Zhang.

Zhang Gaonan: If I have to give an example, "Xinkailai" that we are currently tracking is a classic case of domestic substitution. The key to domestic substitution lies in high - end and sophisticated fields, especially in the bottleneck areas of high - end semiconductor manufacturing processes. Xinkailai was established to break through the bottlenecks in high - end semiconductor manufacturing. It aims to tackle the almost full - stack high - end semiconductor production equipment cluster below 7 nanometers with almost 100% domestic supply chain requirements. This is a typical case of high - end equipment breakthrough at the national level across the entire industry chain. I think it is the most representative microcosm and demonstration of domestic substitution, especially in the semiconductor industry. Thank you.

Zhao Cong: I understand. With the help of the trade war, the semiconductor industry is indeed a very typical field. In the past two years, we can see that the self - sufficiency rate in various dimensions has significantly increased. Now, let's invite Mr. Yang to share.

Yang Yang: Actually, it's similar to what Mr. Zhang said. I think that in the first half of this year, the hottest area of domestic substitution has been reflected in the secondary market. Cambricon is a representative example of the obvious manifestation of domestic substitution this year. In fact, from our communication with enterprises, although products like Ascend and Moore Threads' S - series already have some inference capabilities, including the computing power for models with parameters of around 50 billion or 500 billion. However, in the calculation of trillion - level models, there is still room for improvement. At the same time, our domestic computing power cards still have certain deficiencies in energy - efficiency ratio and need further improvement. For example, while an H100 can complete a task with one kilowatt - hour of electricity, Cambricon may consume 1.5 to 2 kilowatt - hours. I hope to see more improvements and progress in domestic computing power cards in the future. Thank you.

Zhao Cong: I understand that Mr. Yang actually believes that the price is determined by real - money votes. So, from an industry perspective, a moderate and reasonable bubble is actually necessary and beneficial to the industry. Now, let's invite Mr. Tang from Hechuang to give an example.

Tang Zujia: From my observation, domestic substitution has indeed entered the deep - water zone. There are almost no new companies or products that are purely for domestic substitution. Just yesterday, I communicated with an enterprise about their latest product progress. Their product planning is to first design a new - architecture product for the incremental market. Usually, after the sales and market position of this product are stabilized, they will consider designing a Pin2Pin product to replace foreign competitors in the existing market. It can be said that the thinking about domestic substitution has changed greatly compared to the past.

Another point is that, as mentioned earlier, computing power and semiconductor manufacturing have been the common focuses of domestic substitution in recent years, and the competition in the first half has reached a certain stage. Whether from the perspective of stock price and market value or the actual performance, the gap in single - card computing power in the computing power field is not that obvious. Therefore, in the past two years, we have paid more attention to the progress in communication and storage. Since a computing power cluster is a product that combines computing, communication, and storage, compared with computing power, the gap between domestic products and foreign ones is larger in communication and storage. Especially in communication, although Huawei has the 384 product, it is based on the 910 computing power architecture. Given that the 910 is not a GPU architecture, the integration efficiency of the 384 with other domestic GPU products is still not ideal. In recent years, a large proportion of the improvement in NVIDIA's speed and efficiency has relied on the improvement of communication capabilities. From our perspective, we are now more focused on communication.

Therefore, we have made more investments in the communication field this year. There are companies that are replacing NV link and NV switch at the protocol layer, there are also active high - speed copper cable communication products at the physical layer, and optical products with LPO and CPO solutions. With the development of technology, domestic products in China have received more support and cultivation from the application side, and in some niche fields, they have clearly outperformed traditional European and American products.

Zhao Cong: Thank you, Mr. Tang, for your sharing. It also shows us some forward - looking layouts of Hechuang. This reminds me that NVIDIA was able to build a trillion - dollar empire mainly because it acquired Mellanox back then, which solved the problem of communication capabilities. Now, let's invite Mr. Liang.

Liang Feng: I'd like to share some data, from which we can see the progress and challenges of the domestic substitution process in China's semiconductor industry. In a sense, the comprehensive domestic substitution process of China's semiconductor industry started with the "ZTE incident" in 2018. Overall, the domestic substitution rate has increased from 10% in 2018 to 24% in 2024, and it is expected to reach about 26% this year. In different semiconductor sub - fields, the domestic substitution rate has also increased differently. For example, in the storage industry, the domestic substitution rate was almost zero in 2018, and the substitution rate of DRAM increased to 18% in 2024 and is expected to reach 23% this year. For semiconductor equipment, the domestic substitution rate increased from 4% in 2018 to about 13% in 2024 and is expected to exceed 20% this year. That is to say, whether overall or in sub - fields, the substitution rate is gradually increasing, and there has been a rapid increase in different fields. This is one side of the coin.

On the other hand, although domestic substitution has made comprehensive and substantial progress overall, the substitution rate is still very low. Especially in key fields, such as the CPU, the substitution rate increased from 0% in 2018 to 10% in 2024. In many core equipment and high - end chips, overseas companies still dominate and monopolize the market. For example, in the AI field that everyone has been focusing on recently, there are products like Huawei Ascend, Hygon, and Cambricon in the field of computing power chips. However, in the field of high - speed data transmission, high - bandwidth intelligent network card chips that can be compared with NVIDIA's CX series have not entered the large - scale mass - production stage.

Overall, the domestic substitution process of China's semiconductor industry has indeed entered the second half of the deep - water zone. Most of the easy tasks have been completed, and what remains are hard nuts to crack. That is, many core technologies and key fields still need to be broken through. But as the saying goes, although the road is long and arduous, we will reach our destination if we keep moving forward.

Zhao Cong: Thank you, Mr. Liang, for your in - depth and quantitative sharing. We can see that in the first half, the gap for substitution has been opened. From the perspective of industry development trends, once the gap is opened, the substitution rate will definitely increase, and this path is irreversible. Although the road in the second half is still long, from another perspective, it means that our potential market space is still huge, which is also an opportunity for future investments.

Next, we will ask questions one by one. First, I would like to ask Mr. Zhang to share. Mr. Zhang has a background in technology and has gradually shifted to investment. We can see that hard - technology investment actually requires a high level of technical judgment from investors. Since the entire industry is constantly evolving and technological routes are constantly changing, there are various changes in the process. I would like to ask Mr. Zhang to share how investors can continuously maintain advanced technical judgment to avoid misjudging early - stage projects?

Zhang Gaonan: Judging technological trends and success factors is indeed difficult and poses a huge challenge to investors. I think we can improve in the following aspects: First, we should cultivate cross - disciplinary learning abilities. We must not be limited to a single major. I studied automatic control as an undergraduate, system engineering as a postgraduate (I had to cross two disciplines), and then switched to the communication industry for my doctorate. I have been involved in a large amount of software development, hardware design, factory quality management, and Internet big - data projects. I am actually a product of cross - disciplinary education. Cross - disciplinary learning and the integration of knowledge are very important for investors. This is the most basic prerequisite and requirement for successful technology investment. How can we achieve cross - disciplinary learning abilities? First, we must be brave enough to cross - learn and break through boundaries. In addition, most cross - disciplinary learning ultimately boils down to mathematical foundation and ability. The mainstream of large models is the Transformer framework. The core of Transformer is deep learning, and the foundation of deep learning is neural networks. Neural networks mainly involve matrix operations and probability. If you master linear algebra and probability theory well, you will have a much deeper understanding and imagination of the ability boundaries of large models than others. Therefore, we should be brave enough to spend time building a certain mathematical foundation, which will greatly improve our cross - disciplinary learning abilities.

Second, no matter how strong our learning ability is, we should respect the professional thresholds of different fields. It is a good way to actively consult real experts in the industry and academia for the truth. However, please note that we must consult in a critical way. Because not all experts are good at investment. In my personal experience, we can usually consult them about technical difficulties, objective challenges, industrial structure, and other objective information. Because they have more experience and are more accurate. However, we must analyze the trend prediction by ourselves, and their opinions can only be used as a reference. Industry experts usually tend to be more pessimistic about future innovation, and their pessimism often prevents investors from seizing uncertain opportunities. Therefore, we need to strike a balance on our own.

Third, and I think this is the most important point, as an investor, we should have a very high - level systematic thinking ability. No project is perfect, and each has its own advantages and disadvantages. The so - called systematic thinking ability means being able to think out the real underlying causal logic from a large amount of seemingly chaotic information in different dimensions, abstract the essence of things, and make correct predictions. For the same objective fact, different people may draw completely different subjective conclusions. I think an investor's systematic thinking ability determines the upper limit of an investor's excellence. The cultivation of this systematic thinking ability requires both talent and a large amount of post - natal effort. First, we need to have a relatively better talent for abstract thinking than ordinary people, and at the same time, we need to accumulate a large amount of practical experience through observation and learning. I think systematic thinking ability is the most essential.

Fourth, in investment, there are some technical means in the terms. When facing projects with particularly high technical challenges and high uncertainty, we may need to make some targeted protections in the terms and agreements, such as the completion of phased goals, the rhythm of payment, and special financial bindings. I think this is a technique rather than the essence, but it can help reduce project risks to some extent. Thank you.

Zhao Cong: Mr. Zhang has shared a lot. I understand that it is actually a very difficult thing. We need to have a certain foundation of knowledge and thinking framework and continuously learn. So, it seems that investment is actually a hard - working