The ongoing digital economy wave is sweeping the globe, driving a qualitative transformation in the "connected ecosystem." From the era of PC internet connecting people, to the mobile internet era enabling constant connectivity, and now to the AI-driven intelligence of all things, the fundamental logic of the internet is being rewritten. This profound change is not only reshaping industrial structures but also significantly raising the cognitive barriers and decision-making complexity for technology investments.
Huà'ān Fund, as one of the earliest asset management companies to deeply cultivate technology investments, has assembled the "Huà'ān Fund Technology Alliance," which combines industrial-academic backgrounds with investment research strength. Leveraging deep industrial expertise and team collaboration, the alliance continuously expands, deepens, and iterates its exploration within the technology investment domain.
Hu Yibin, Co-Chief Equity Investment Officer at Huà'ān Fund and a pioneer in high-growth sectors, transitioned from a technology company to the Huà'ān investment research team. Over a decade, he has focused extensively on sectors like computing, electronics, and media, closely monitoring disruptive opportunities emerging across various segments of the AI and digital economy supply chains. Since officially joining Huà'ān Fund in May 2015, Hu Yibin has progressively broadened his expertise beyond TMT and tech growth stocks, evolving his investment framework to become a "high-growth capturer" focused on growth investing with a balanced style.
This first dialogue with Fund Manager Hu Yibin shares his key insights below.
1. Internet Still in Early Classical Stage, Embodied AI Awaits Breakthroughs Moderator: Our discussion today focuses on insights into the internet ecosystem and uncovering new investment paradigms. What stage do you believe the industry is currently at? Is it a period of stock game after peak traffic, or at a new starting point for value reshaping? Hu Yibin: Many believe that from PC to mobile internet, traffic has saturated and monetization is near its limit, suggesting the industry is in a stock game phase post-peak traffic. However, if we consider intelligentization, the Internet of Things, and an internet capable of emergent thinking, the internet industry remains a truly vast frontier and is at a new starting point for value reshaping. In this vast frontier, every star is a sun – in the era of interconnected things, every connected entity can become a nascent internet industry.
Where will the next phase of the internet lead? We may witness space-air internet, the blending of physical and virtual worlds, and intelligent agents becoming increasingly smarter. This is my understanding of the internet's future vast frontier, while the business models we see today – e-commerce, gaming, advertising – merely represent the boundaries of the previous era.
In the future, every intelligent agent will become our tool, potentially even exhibiting emergent thinking, much like each sun illuminating a specific vertical field. Therefore, the current internet is still in a very early classical stage.
Moderator: A hot topic recently is the combination of AI large models and embodied AI. Many consider this combination the potential ultimate form of next-generation intelligent agents. What is your view? Hu Yibin: The combination of intelligent agents and large models can indeed equip physical entities with intelligence and emotional quotient, achieving unity of knowledge and action. However, there is still a long way to go before achieving the embodied intelligence truly needed by humanity.
The embodied intelligence I understand requires not only unity of knowledge and action but also self-awareness, self-actualization, and the ability for intelligent agents to produce actions in the physical world – this is the form most needed by humans. Similar to the carbon-silicon endpoint depicted in "The Matrix," discussing this topic today feels more tangible. A true intelligent agent should possess limbs, the capacity for action, and an emotional quotient surpassing humans – meaning the ability to gain insight into its core and self-perception.
Currently, intelligent agents exceed expectations in knowledge base construction and emergent capabilities but still fall short of genuine emotional connection and self-actualization. Components in the execution phase, such as motors and lead screws, are advancing rapidly, but achieving human-required functionality will likely take at least 3-5 years. Nevertheless, I believe both the intelligent agent industry and the execution aspects of embodied AI have substantial room for development, presenting numerous investment opportunities along the way.
2. AI + Internet Reshapes Industrial Structure, Pay-Per-Use and Value-Added Services May Become Mainstream Moderator: Bringing the perspective back to the industry level, which segments is AI + internet currently reshaping? Hu Yibin: We can look at three layers. First is the foundational layer: computing power. This includes not only chips, IDCs, and connectivity components but, more importantly, capital behind them – in the medium to long term, computing power will replace traditional factors of production, making capital's role essential. Second is the model layer. From the "hundred-model battle" of 2023, characterized by free investment and competition among major players, to today where large providers are achieving operating cash flow through subscriptions and token revenues, this differs from the 2000 internet era, as returns follow investment more quickly. Competition among model providers is trending towards oligopoly, but it may not necessarily result in one or two winners taking all, as differentiation persists across many verticals, offering good opportunities for discovery. Third is the application layer. As computing power and models mature and become accessible, a multitude of application developers will emerge. Embodied AI, autonomous driving, and others represent new application opportunities. Looking back, we realize how limited our past understanding was. In the future, across fields like healthcare, education, law, and finance, each application point is a star – and each star is a sun, releasing immense commercial potential.
Moderator: Which business models within the internet industry are you more optimistic about? Hu Yibin: In the future intelligent internet era, business models will see more novel explorations: shifting from time-based billing to traffic-based billing, and further to pay-per-use, representing monumental changes. For example, the gaming business model has evolved from one-time purchase or time-based fees towards pay-per-use for value-added services – users can enjoy basic services even without payment. In the future, value-added services and subscription models will become increasingly important, and user behavior will likely shift significantly towards pay-per-use. There are two primary targets for these payments: first, computing power infrastructure providers (cloud platforms), whose pay-per-use models are well-established and difficult to disrupt long-term; second, model providers, whose business models will diversify – free offerings, one-time purchases, knowledge payments, time-based subscriptions, and token consumption fees will likely coexist. While we currently appreciate the cash flow returns from subscription models, as intelligence scales, pay-per-use and value-added services may become the new mainstream.
3. Product Manager Mindset × Human Warmth: Building Moats in Active Management Moderator: How has your product manager background influenced your investment perspective? What non-financial metrics do you focus on when assessing a company's value? Hu Yibin: My industry experience constantly reminds me of three things: First, continuous learning. The lifeblood of growth stock investing is learning. Only by remaining humble and continually exploring the unknown can one stay on the right track. Second, the user is king. Behind investing are hundreds of thousands, even millions, of fund unitholders. I manage the fund like operating a product: besides pursuing excess returns, it's crucial to help unitholders persevere or avoid risks when needed. Third, maintain a sense of crisis. Active fund managers face challenges from AI-driven quant and index investing. Our survival hinges on leveraging human advantages over AI: firstly, not simply extrapolating the future from the past, but having the imagination and creativity to break boundaries, treating investment as a work of art aiming for 90/100; secondly, applying "human understanding" to industrial decisions, assigning different weights of faith and information to various entrepreneurs – this is a warmth, an emotional intelligence, that AI cannot yet reach. Regarding non-financial metrics, the core lies in the warmth AI cannot perceive from human connections. Purely rational, quantitative investing is easily replaceable. Combining objective professionalism with this intuitive faith in people and matters, creating a spark of soulful connection, might be the foundation for long-term success in active investing.
Moderator: Having been with Huà'ān Fund for over ten years, how have you and the platform mutually enabled each other's success? Hu Yibin: First is the professional environment characterized by tolerance and trust. Second is the investment research culture of selfless sharing. Huà'ān is like a family; while everyone has their areas of expertise, whenever opportunities or risks are identified, they are willingly shared and discussed. Working alone might yield a result of 60/100, but uniting together can deliver an outcome of 90/100. Third is the platform's strong systemic engineering capability. When we encounter knowledge gaps, the platform can quickly address shortcomings through talent acquisition, resource allocation, and process optimization. It is this culture of inclusive trust and harmonious coexistence that has supported my transition and evolution from a sector-specific fund manager to a whole-market manager, and from managing small products to serving millions of unitholders.
Comments