Zero2IPO Investment Research Institute | HyT Capital's Sun Yelin: "Tracks, Logic, and Investment Opportunities" at the Forefront of the Wave
As the internet era gradually fades into the distance, hard tech investment has stepped into the spotlight, drawing significant attention.
With the deep overlay of strong technology cycles and the growing trend of industry specialization, this sector demands far more from investors than just "pure finance." It requires not only higher strategic vision and faster decision-making, but also a deep understanding of technology and a keen insight into how technology translates into commercial value.
There is a category of investment firms that "come from the industry" and "go back into the industry," demonstrating strong capabilities and vitality in the hard tech investment era. Hua Capital is one such representative.
Hua Capital is a VC firm with a strong industry background, deeply focused on investing in intelligent technology and the industry transformation it drives. Since its founding nine years ago, the firm's AUM has exceeded RMB 7 billion. Nearly 70% of its portfolio companies have received follow-on financing from top industry players and well-known investment institutions, and over one-third have grown into unicorns or potential unicorns. The recent LP re-investment rate has also exceeded 70%.
What insights do they have on intelligent technology investing? Which niche sectors are they closely watching? How do this new generation of entrepreneur-investors capture these investment opportunities?
This article is an edited digest of the publicly shareable content from a 50,000-word speech delivered by Sun Yelin, Founding Partner of Hua Capital and Mentor at the Zero2IPO Investment Research Institute, sharing his successful experiences in intelligent technology sector selection.
Former Senior Vice President of a global leading technology company, with extensive experience in innovative technology product R&D, global regional management, technology corporate strategy and organizational management, investment decision-making and management, as well as rich industry resources. Representative investment cases include: InnoScience, NuVolta Technologies, Silicon Integrated, Vertilite, JSCJ, GalaxyCore Inc. (SH688728), FLAGCHIP, Shanghai Visinex Technologies Co., Ltd., JOYWELLSEMI, Shenzhen Jaguar Microsystems Co., Ltd., Tuoyi Technology, Kema Technology (SZ301611), Sandtek Semiconductor Technology (Shanghai) Co., Ltd., Jingwei Hengrun (SH688326), Wuhu Atech Automotive Co., Ltd., Hanshow Technology, Shenzhen Ldrobot Co., Ltd., KeenData (Beijing Kejie Technology Co., Ltd.), and others.
Hua Capital was founded in 2015, focusing on investing in potential industry leaders in the intelligent technology sector, with key investments in semiconductors and artificial intelligence, covering the entire semiconductor value chain, AI infrastructure, intelligent terminals, and industrial intelligence. Hua Capital's RMB and USD funds have a combined AUM of over RMB 7 billion, with capital primarily entrusted from well-known national guidance funds, local guidance funds, market-oriented Fund of Funds (FoF), financial institutions (banks, insurance, securities, etc.), industrial capital, real estate groups, family offices, and university endowments.
I have nearly 20 years of experience in the industry and have been investing for 9 years now. My background shares much in common with many alumni of Shaqiu, so there is a lot of resonance. At the time, I thought the transition from industry to investing would be easy, because industry requires deep engagement on the front lines, while investing does not involve operational intervention. But eight years later, my feeling is that investing is much harder than running a business.
In industry, you are optimizing within a defined field and direction. In investing, you must make numerous choices across a vast and complex landscape, and after making those choices, outcomes are often difficult to control. As a result, I have developed a deep respect for investing. Investors need to engage in lifelong learning — it is an endless journey.
Looking at Hua Capital's nine-year journey, there have been both achievements and lessons. We have engaged in many practical explorations in the VC space. Today, I would like to share some of the experience we have gained from our successes, as well as some insights we have derived from our setbacks — all centered around the intelligent technology sector.
01
Sector Selection · All In on Intelligent Technology
I am often asked what qualities make a good VC investor. I have summarized the following three:
First, imagination. VC investments are often in areas that do not yet exist in the world. Without imagination, it is hard to excel in this field. Second, deep thinking capability. One must look at industries, trends, sectors, enterprises, and founders — with a thinking style that is profound and goes to the essence. We have invested in nearly 100 companies and have come to deeply feel that a company's success depends on too many factors. Only through deep thinking can one grasp them. Third, execution capability. VC investments often involve highly asymmetric information, requiring strong networking and project execution skills to reach better entrepreneurs and top industry experts, and to stay ahead of cutting-edge technology trends.
In short, these three qualities are key to excelling in VC investing. However, in reality, it is almost impossible to find someone who possesses all three perfectly — which is why team collaboration is also very important.
Technology Has Innate Vitality
Since its third year of operation, Hua Capital has focused on the intelligent technology sector, which we prefer to define as "third-generation information technology." It did not emerge out of thin air, but rather developed from the foundation of the previous two generations of information technology. The first generation began in the early 1970s with the birth of the PC. This innovation drove nearly half a century of rapid global technological development, and the extraordinary rise of the US economy can also be traced back to this period.
Entering the mid-1990s, the second generation of information technology emerged. During this period, PCs became interconnected through optical communications, routers, and switches, enabling data sharing. Sitting in front of a computer, we could access information from around the world. From the massive production of digital information starting in the 1970s, to connecting it all together in the 1990s, information technology made a leap forward.
Entering the 21st century, these interconnected data began to demonstrate intelligent potential. In 2016, the first generation of the AI revolution — computer vision technology based on neural network principles — achieved breakthroughs. Then, in 2022, the breakthrough of models based on the Transformer architecture marked the arrival of a new AI era.
When examining investment opportunities in each generation of information technology, we often find that when people mention AI investing, the first thing that comes to mind is models and their applications. But this is only the tip of the iceberg. To more systematically analyze the third-generation information technology or intelligent technology sector, we adhere to a four-dimensional perspective: semiconductors, infrastructure, intelligent terminals, and intelligent applications. These four dimensions not only help us focus our efforts, but also broaden our horizon, making it easier to capture potential investment opportunities.
The "Four-Layer Information Technology Cycle Structure" that Hua Capital has developed around these four dimensions has become our fundamental analytical framework. Under this framework, we can more clearly recognize where our strengths lie — which opportunities are worth pursuing, which may need to be deprioritized or receive less capital, and in areas where we have particular expertise, we need to increase our commitment.
The development of this framework is grounded in a deep context, which is closely tied to the evolution of the global economy. We reference the global GDP growth trend chart starting from 1970. What stands out is that from the dawn of human civilization to the 1970s, the total GDP accumulated over that entire period is almost negligible compared to today's GDP. This means that despite thousands of years of human toil and a splendid history of civilization, the economic output before the Industrial Revolution appears almost insignificant.
Today, the economic prosperity we witness stems primarily from industrial civilization and technological civilization — especially the new wave of technological revolution driven by information technology. These transformations have dramatically increased global economic output.
Why has the emergence of information technology brought about such enormous economic growth? One notable reason is the substantial improvement in efficiency. From machinery to electricity, and then to information technology, every technological revolution has brought leaps in production efficiency. Information technology, in particular, with its rapid information flow and efficient resource development, has become the most important force driving economic growth. Since then, the global economy has continued to grow, with new markets such as the digital economy and the attention economy constantly emerging. Around these emerging markets, entirely new industrial clusters have formed — a phenomenon that never occurred in the thousands of years of human civilization before.
When examining the challenges currently facing China's economy, we can adopt a multi-dimensional perspective: political, economic, capital, and technological. The governance system is the most fundamental underlying logic of development. These elements are undoubtedly interconnected, but they do not move in perfect synchronization over time. The strengths or weaknesses of the political system and governance structure do not immediately reflect in the economy, capital, or technology — there is a certain lag before their effects become significant. Similarly, there is a temporal disconnect between the economy and the capital markets. Over the past three decades, China's economy has continued to grow at a high speed, but the stock market has remained relatively stable, with occasional fluctuations, and its overall trend has not been fully aligned with economic growth. The same applies to technology. Taking the first new technological revolution starting in 1970 as an example, its impact did not begin to significantly reflect in the US economy and capital markets until the 1980s.
The Competition and Cooperation of "China-US" Technology
In the current global economic landscape, China and the United States are undoubtedly the two major forces in intelligent technology. This was fully reflected in an internal speech by Eric Emerson Schmidt, former CEO of Google. In the speech, he shared his views on intelligent technology, stating that countries capable of competing in this field need to have sufficient capital, a strong talent pool, a robust education system, and the will to win — with China and the US being the two primary nations. This speech offers high reference value for understanding current global technology trends and is worth reading.
When examining the differences in technological development between China and the US, we need to conduct in-depth analysis from multiple dimensions — such as business models, talent reserves, capital characteristics, computing power, and algorithms — and engage in comparative studies. This helps us make more accurate judgments. These dimensions are dynamic, including governance structures, all of which are constantly evolving. Therefore, we continuously monitor and evaluate these dimensions.
For example, in terms of innovation models, China and the US exhibit significant differences. American culture encourages original innovation, with most startups focused on 0-to-1 breakthroughs. In China, we excel more in engineering innovation — making large-scale investments and optimizations based on existing ideas. This difference is especially evident in areas such as electric vehicles and photovoltaics, where most of the original ideas originated in the US, but rapid development has been achieved through China's engineering innovation.
In terms of talent base, the US has an efficient system and substantial investment in frontier R&D, attracting many doctoral students from top universities to apply for research positions. This close collaboration between universities and major tech companies drives breakthroughs in cutting-edge technologies. In China, we have a large number of dedicated frontline engineering talents, but the talent structure for frontier R&D and innovation still needs improvement.
The differences in industrial foundations between China and the US are also clear. China has advantages in large-scale advanced manufacturing, while the US leads in software and digitalization. As we have seen, these differences have led to divergent development trajectories in areas such as SaaS, with investments in related fields also showing distinctly different patterns.
Although China and the US differ in these aspects, I believe this does not mean there are no investment opportunities. On the contrary, precisely because of these differences and ongoing changes — whether in semiconductors, terminals, or applications — we can see potential opportunities shimmering on the horizon.
From a technological perspective, the variables in technology over the next ten years will be even greater. As a result, we have adjusted our investment strategy — reducing the number and amount of investments, while increasing research time and the number of projects we evaluate. We believe that in the coming decade, whether in China or the US, there will be more technological innovations and investment opportunities emerging.
Take AI for Science as an example. Although this field may sound more like a US strength, preliminary research suggests that China could also achieve breakthroughs in this area. This is because AI for Science is primarily applied in pharmaceuticals and materials, and China has advantages in industries such as batteries and photovoltaics, with large amounts of accumulated data — these industries are exactly the key application scenarios for new materials.
Furthermore, with the proliferation of AI tools, the productivity of the world's brightest minds will be greatly enhanced, and the output and commercialization of scientific and technological achievements will accelerate. This is a very inevitable trend.
Of course, we also understand the difficulties and challenges entrepreneurs have faced in recent years. As I mentioned earlier, the economy, technology, and capital are coupled — but they are not fully synchronized at the same moment. Therefore, when we see the upward force of technology, we should maintain confidence and seize investment opportunities.
02
Investment Logic · Insight into Transformation Through a Focus on Fundamentals
We often ask ourselves: What does AI create? What does AI strengthen? What does AI disrupt?
Before answering these questions, a more fundamental one requires our consideration: What is the true nature of this wave of AI technology? Only by accurately understanding the nature of this AI wave can we better comprehend and fill in the three scenarios above.
The primary change in this wave of AI is not merely an upgrade of traditional architectures, but rather a major optimization of neural network architectures. Take Transformer as an example — it is a key innovation in neural network architecture, with the multi-head attention mechanism at its core. For investors, understanding its basic principles is quite necessary, as it helps us grasp the boundaries and core essence of AI technology, thereby enabling us to judge what AI can and cannot do.
The brilliance of the Transformer architecture lies in its ability to enable large-scale parallel computing, which makes training ultra-large-scale models possible. During the training phase, through self-supervised learning, the model can learn vector correlations without human intervention and store these correlations within the model. It is like a knowledge compressor, converting text or phrases into vectors and determining the correlations between them. As the parameter scale increases, the model's resolution of vector correlations improves, thereby demonstrating stronger intelligence.
However, it is important to note that this intelligence is based on correlational reasoning, not causal reasoning. AI does not know which is the cause and which is the effect — it simply predicts the next most likely word or phrase based on the correlations it has learned. Therefore, although AI demonstrates strong reasoning capabilities in some areas, this reasoning ability remains limited.
Once we understand the core essence of AI, we can more clearly see the boundaries of its applications and more accurately predict the direction of the next technological revolution and the transformations it may bring. Just as airplanes, although inspired by the flight principles of birds, have far surpassed birds' flying capabilities, the Transformer model has also far surpassed the capabilities of the human brain in some respects. Even with limitations such as high power consumption, we must acknowledge that the continued development of AI will bring disruptive opportunities and challenges.
Creation, Strengthening, and Disruption
When we deeply understand these concepts, we can more accurately determine which areas are being created by AI, which are being strengthened, and which are being disrupted.
We have noted that Eric Emerson Schmidt mentioned in his speech that he has already invested in next-generation systems designed to optimize the Transformer model — an area currently drawing close attention from both academia and industry.
Based on this learning capability — and particularly where it surpasses the human brain — lies in the reproducibility of knowledge. Once learned, the compressed knowledge can be replicated. Therefore, a large number of correlation-based tasks can be performed by AI, including knowledge and civilization that have already emerged in human history. Once we understand these technical fundamentals, we can infer what kinds of "new species" will emerge.
▷ Based on our research, one of the most anticipated innovations is the "AI assistant." In the future, each of us will likely have such an intelligent assistant to help handle daily tasks.
▷ The vast array of enterprise software used on the business side may also disappear in the future, replaced by digital robots that can coordinate various professional functions — HR, sales, service, fault diagnosis, and more.
▷ Hardware-related areas are also very popular. Many companies that have only been in business for a year or two, without even having a physical robot product yet, have seen their valuations soar. This reflects a broad consensus on the trend, but truly AI-native intelligent agents will still take some time to materialize.
▷ Digital twin factories have been discussed for a long time, but have been difficult to realize. Because factory systems prioritize reliability, have low willingness to change, and adopt new technologies slowly. However, with this wave of AI and 5G, this area may be brought back onto the agenda.
▷ At the infrastructure level, we firmly believe in the continuous evolution of algorithmic models. The Transformer needs optimization and enhancement, and more disruptive technologies may also emerge. The industry is gradually exploring new research directions, but challenges in supporting computing chips make these directions difficult to implement.
▷ There are also many new opportunities on the semiconductor side. The imperfection of the current Transformer model is that every operation requires activating all operators, whereas the human brain does not activate all neurons simultaneously when thinking. As a result, AI requires new forms of heterogeneous computing, where computing and memory need to be more closely coupled.
For VC investing, once we foresee these trends, we inevitably begin to fear "missing out on good projects." A great project may already be emerging in some corner of the world, but we have not yet reached it. Imagination, deep thinking, and execution capability need to be connected in order to bear fruit. Currently, our greatest anxiety is the fear of missing out.
In areas being strengthened by AI, smartphones are a typical example. We have invested in several upstream semiconductor companies in the consumer electronics space. They were somewhat pessimistic before due to the lengthening replacement cycles, but I believe a new wave of "upgrade demand" will soon arrive, because the emergence of AI phones is highly probable. An AI phone is not simply about installing a ChatGPT app — it requires greater computing power and storage capacity. As a result, mobile phone SoCs and related components will see a new generation of upgrades. It is also possible that new phone brands will emerge around AI phones.
In areas being disrupted by AI, complex businesses that rely on human delivery may be among them — particularly those with low accuracy requirements. One area worth discussing is e-commerce, which may also be disrupted. In the future, e-commerce platforms may serve AI, which in turn serves humans. A shift in the customer of service will lead to a fundamental change in the nature of the business, and existing advantages may turn into disadvantages.
03
Opportunity Capture · Upholding Vision and Aligning Knowledge with Action
Earlier, we emphasized the importance of imagination and deep thinking. However, in the VC field, possessing these abilities alone is not enough to directly generate investment outcomes. The key lies in how to translate these thoughts into concrete action — that is, to identify and seize investment opportunities, deploy capital at the right time, and then exit at the right time as the company grows.
To achieve this, going from cognition to action requires a complete process. First, we need to have a clear vision of the future. In VC investing, blindly following the crowd is not advisable. Simply following a trend or investing because a major institution has participated is a dangerous approach.
Successful investment requires independent judgment, which stems from deep insight and a strong vision of the future.
Vision is not daydreaming — it is built on a deep understanding of the fundamentals of technology and accurate predictions of technological trends. We need to think about how technology will integrate with business scenarios, which requires deep understanding and reflection on both the technology and business sides.
The future vision I believe in is: "Twenty years from now, the world will present itself as a digital twin." Most physical entities will be connected to the cloud through sensors and wired/wireless networks. The cloud will have central and edge computing power, as well as highly intelligent next-generation models — which may possess genuine reasoning capabilities — enabling the physical world to exhibit high levels of intelligence.
Under this vision, we focus particularly on investment opportunities with a 3-to-5-year horizon. In searching for such opportunities, we encounter two types of seed projects: causal and emergent. Causal projects are those where the drivers and outcomes can be clearly identified — such as domestic substitution in semiconductors. Emergent projects are characterized by unpredictability — such as the next-generation AI assistant, where we don't know where it will emerge or who will create it. For causal projects, we conduct in-depth research and analysis. For emergent projects, we need to diligently scan the landscape and talent pools to discover new projects and people.
Who Will Win?
Many outstanding entrepreneurs are interested in personally venturing into the investment field or building teams to conduct investment activities. I would also like to share some relevant perspectives and methods, in the hope of offering some inspiration. These insights are the result of years of continuous iteration, repeated reflection, and many lessons learned from failures.
First, motivation. Curiosity, passion, perseverance, and vision are all indispensable motivational factors — because the VC industry demands the highest standards. After all, we are investing in unknown territories. What gives us the conviction that a particular company will surely grow and succeed in the future? Although we can make judgments based on logic, there is at least a 50% or more chance that uncertainty will arise over the next five to ten years — policy changes and various unpredictable events. We must accept these uncertainties with equanimity. What we can truly grasp, in the end, is the logical part; the rest relies on this "motivation."
As my understanding of investing has deepened over time, I have gradually come to appreciate what truly defines a good investor. In fact, it is not just true for investing — the same applies to being an entrepreneur. What matters most is motivation — whether you truly love this profession — loving to invest in the technology of the future. Being full of curiosity is equally crucial. Without this passion and curiosity, you cannot form a relatively clear vision of the next five to ten years, or ten to twenty years.
Second, cognition. Cognition is even more challenging, because the technological variables over the next ten years will be greater and faster. The constant evolution of technology requires us to conduct in-depth research. What is common knowledge? What are the underlying patterns? What are the driving factors behind them? ... All of these need to be thoroughly understood. For example, if you invest in AI without understanding the Transformer model, that is blind investing. Similarly, if you invest in embodied robots without understanding basic concepts such as data sources, simulators, reinforcement learning, imitation learning, and end-to-end models, that is also unwise — because you have no idea whether the company you are investing in will be quickly overtaken by new technological approaches in the future.
Third, capability. This primarily involves the two fundamental abilities: induction and deduction. In one article, Yang Chen-Ning brilliantly articulated these two abilities. He noted that the education he received in China emphasized deductive ability — once you know a formula or a principle, you can continuously apply it. Later, he discovered in the US that American education placed more emphasis on induction — observing phenomena and then summarizing the underlying patterns. Both induction and deduction are vital in the scientific community, and they are equally important for us in VC investing. We need both inductive and deductive abilities. By interviewing numerous technical experts and entrepreneurs, we continuously induct and synthesize, discovering the essence and patterns, and then — based on these core insights — we can infer future trends, which is deduction.
Although the investment work we do is not scientific research, our way of thinking must align with that of scientists. Most of us Chinese have been educated in deductive thinking — learning a formula and quickly solving problems. But in VC investing, deductive thinking alone is not enough. We also need inductive thinking to cope with the challenges of constantly evolving technological variables.
Finally, action. To reach top scientists, industry leaders, and experts across various functions, and to obtain comprehensive information, we need network maps, project maps, and competitive maps of major enterprises. Of course, no investor or institution can achieve perfection in this regard — this is merely our standard and goal. Since drawing these maps requires a great deal of time and effort, we must be "targeted" in our approach, selectively focusing on certain areas and sectors.
Once we have the above model in place, how do we determine "who will win" in investment practice? We adopt a methodology for research that Ray Dalio of Bridgewater Associates mentioned in his book Principles — a methodology for decision-making and thinking. I believe this methodology is well-suited for investment and research thinking, and it helps us put induction and deduction into practice.
The methodology is roughly as follows: when thinking about anything — for example, how to invest in humanoid robots — a horizontal structure should quickly form in your mind, including elements such as technology, competition, costs, and the founder. Taking technology as an example, you need to vertically break it down into macro, meso, and micro levels. At the macro level, quickly form a structure to understand the technological roadmap. At the meso level, break down the technical challenges, and continue breaking down layer by layer. Once the breakdown is complete, faced with a vast amount of information points, you identify the most critical information through induction. Based on that information, you then apply deduction to think about what scenarios are applicable, what types of founders are more likely to succeed, and what types of companies to prioritize for investment.
However, the unfortunate reality is that the vast majority of us have not received this kind of rigorous thinking training. As a result, common mistakes include: being overly sensitive to information points one is most familiar with and overestimating their impact, leading to misjudgments of the entire sector; or analyzing deeply in one dimension while having significant structural flaws at the overall level, which can also lead to misjudgments.
In our exploration of technology investing over the years, we have accumulated some valuable cognitive assets and have also been through the baptism of both successes and failures. We are still far from responding to the challenges of our time with satisfactory tangible results. Knowing is easier than doing — from building cognition to closing the loop in action usually takes a long cycle. We look forward to exploring and honing our approach together with partners and peers in the future, contributing to the germination, growth, and expansion of more technology enterprises. END
The above article is republished from the "Zero2IPO Shaqiu Investment Research" 【清科沙丘投研】WeChat public account.
Back
Next article
