Zhang Chuanjie: AI Investment Enters a Critical Validation Period

Ask AI · How does AI investment verification period test market expectations?

“Xingquan Science+”

Investing is both a rigorous science and a philosophical insight; it combines the warmth of long-termism with the art of rational decision-making. Each episode features 10 questions, in-depth conversations with an outstanding Xingquan fund manager, to dissect their core investment strategies and styles, share the latest market views, and provide deep reflections.

“Infinite Frontiers” is the name Zhang Chuanjie gave to his investment portfolio, reflecting his desire to continuously explore the cutting edge of the world. As a rising fund manager at Xingzheng Global Fund with a background in TMT research, he is sharp yet cautious. Ten questions bring you closer to fund manager Zhang Chuanjie.

********************Q1: Please briefly introduce your investment framework.

Zhang Chuanjie: I aim for simplicity, ideally as concise as a mathematical formula. My investment framework revolves around three core variables: demand, supply, and valuation. I believe methodology is universal; however, each industry has different supply-demand structures, valuation methods, and environments.

Based on this, my investment goal is to “buy good things at cheap prices.” “Good things” refer to scarce, high-demand assets, and “cheap” is not about static PE or PB ratios being low, but about being significantly undervalued or having huge appreciation potential when viewed from the future.

********************Q2: As an investor rooted in TMT research since the beginning, how do you understand technology investing?

Zhang Chuanjie: One major realization is that doing well in tech investing is very challenging. From my framework’s three elements:

First, demand for technology is nonlinear and unstable. The ability of tech stocks to generate phased excess returns largely depends on unpredictable demand surges, which can suddenly explode in the short term, causing a strong imbalance between supply and demand within the industry, leading to significant short-term stock price elasticity. This is different from relatively stable demand in consumer goods, making it hard to grasp and predict.

Second, supply is also very complex, not only because many players are involved and competition is fierce, but also due to potential technological shifts. The emergence of a new technological route can cause a “dimensionality reduction attack,” leading to the demise of traditional technologies. For example, currently popular optical modules, market concerns about CPO (co-packaged optics) impact, or new materials like gallium nitride replacing indium phosphide in lasers—these materials can substitute each other, adding to supply-side unpredictability.

Third, valuation is unpredictable. Traditional industries like dividend assets have relatively stable business models, focus on dividend yields, and possess stable “valuation anchors” and bond-like asset attributes. But tech stocks can have very wide valuation gaps; some are not yet reflected in earnings reports but already carry high valuations.

The difficulty in tech investing lies in that pricing rarely depends on clear historical or current fundamentals; most are focused on an “unknown future.” This demands high levels of cognitive depth, trading rhythm, and risk control.

To handle this complex system, my strategy involves two steps: first, acknowledge the complexity; second, actively try to simplify. Under the premise of recognizing complexity, I simplify problems as much as possible. I give up on issues that are too complex or incomprehensible. The tech sector is broad; I seek areas within my capability circle that I can see through, then position accordingly.

********************Q3: What successful investments have you made in the past?

Zhang Chuanjie: Take the recent optical fiber industry as an example. As an industry with long-term oversupply adjustments, fiber optics prices started to reverse in Q4 2025, driven by demand from military drones and AI.

• Demand side: The market initially worried that demand would decline after the war ended. I observed that “security” has become a top priority globally, and military demand will be more persistent than market expectations, even more rigid than AI investments. Fiber optics are shifting from a small market dominated by operators worth tens of millions of USD to a strategic resource worth hundreds of billions, with demand expectations differing from previous views.

• Supply side: The domestic fiber optic downtrend lasted for 7 years, causing prolonged pain. Leading manufacturers were cautious about expanding capacity, and since this demand was overseas, they were reluctant to blindly increase supply, resulting in a supply-demand gap far exceeding expectations.

• Valuation side: Previously, the fiber industry peaked during the 5G cycle, with demand hard to predict, and valuations were low. But driven by military and AI, future demand and innovation prospects have improved, and supply constraints are more severe than expected, creating potential for revaluation.

In retrospect, domestic fiber prices have surged over 4 times in the past three months, and the fiber sector has experienced a strong rally. I will continue to explore undervalued supply-demand mismatches.

********************Q4: What is your approach to portfolio allocation?

Zhang Chuanjie: My allocation mainly revolves around two asset types: demand-driven assets, currently represented by AI-related sectors with rapid supply and demand growth, where demand growth outpaces supply; and supply-driven assets, where demand growth is slow but supply faces clear constraints, such as cyclical sectors under the dual-carbon policy, like chemicals. My allocation is based on my three-element investment framework.

Currently, the portfolio is primarily “technology growth-oriented with some cyclical value assets.” The framework is unlikely to change, but the specific targets may evolve with the times. I prefer to stay adaptable rather than rigidly fixed on labels.

********************Q5: Which industries or sectors do you currently favor? Why?

Zhang Chuanjie: Currently, I focus on: overseas computing power, domestic computing power, and energy security.

• Overseas computing power: I am optimistic about new technologies like cabinet-based optical interconnects, and explosive demand driven by CPUs, NPUs, etc. Chinese manufacturers with global competitiveness and product scarcity are expected to benefit significantly from the AI wave.

• Domestic computing power: With major internet giants investing heavily, I favor infrastructure-related data centers and computing leasing companies. Although performance realization remains to be seen, these resources are extremely scarce and flexible. These assets mainly serve the domestic market and are less affected by overseas supply chains, making them good carriers for domestic computing power.

• Energy security: Recent changes brought by the US-Iran conflicts. Energy is a core resource for national development. Countries will need to rethink their energy strategies, bringing many new changes and opportunities.

********************Q6: How would you describe your investment style?

Zhang Chuanjie: Although I come from TMT research, I do not adhere to a single style. Historically, no style can be sustained forever. When a style becomes mainstream and crowded, expected returns tend to compress, and risks increase. This is common in cyclical sectors like technology, where over-enthusiasm leads to overcapacity and collapse. The same applies to finance. I believe “adversity breeds caution” and always consider risks.

Market labels like growth or value, and market cap categories (large/mid/small), are just results, not prerequisites. I consider liquidity and risk-adjusted returns. For small caps with poor liquidity, I require higher expected returns to compensate. The same applies to allocations in A-shares and Hong Kong stocks. Due to the margin lock-in and exchange rate risks in the Hong Kong Stock Connect, I set higher return expectations for related targets.

********************Q7: How do you view concentration and turnover?

Zhang Chuanjie: My core holdings are limited; I prioritize depth over breadth. For opportunities I am confident in, I am willing to concentrate. I sell when consensus is reached, buy when there are divergences. As Drunkenmiller said: “The market doesn’t care how many times you’re right, only how much you make when you’re right.”

Turnover is a result, not a goal. I don’t predefine holding periods. I sell based on three situations: first, when the recognition gap disappears and market prices fully reflect expectations; second, if I am wrong, I cut losses promptly; third, if better opportunities arise with higher expected returns and cost-effectiveness, I will rebalance.

I consider holding long-term in two cases: one, when a company is doing something difficult but right, and its value remains undervalued; two, when there is a top-down systemic undervaluation opportunity, such as China’s decades-long growth potential. Long-term holding requires deep expertise and continuous learning, which is very challenging but rewarding.

********************Q8: What risk control measures do you implement?

Zhang Chuanjie: They fall into objective conditions and subjective discipline.

Objectively, I adhere to limits on single industry and stock allocations set by the company to prevent concentration risk; in extreme uncertainty, I consider reducing equity exposure to control volatility. Subjectively, if I realize I am wrong, I cut losses early to prevent further damage. Regarding potential style shifts, I prefer to respond rather than preemptively hedge. I monitor core variables (supply, demand, valuation); if systematic, trend-based reversals occur, I adjust positions accordingly.

********************Q9: What is your outlook for the market?

Zhang Chuanjie: First, looking ahead to 2026, the AI industry chain has moved beyond the early explosive phase of 2023-2024, and since 2025, it has entered a mid-term stage. The market focus has shifted from “believing in demand” to “verifying performance,” i.e., whether companies can continue to deliver beyond expectations to justify current valuations. This transition makes tech investing more challenging this year, requiring careful validation.

Second, market pricing factors are becoming more complex. Beyond industry fundamentals, macro variables like geopolitical conflicts and interest rate changes are reshaping short-term valuations. Continuous vigilance is needed. Despite macro complexities, there are still structural micro opportunities. Currently, I focus on two types of strong buyers acquiring scarce resources: one, government considerations for military preparedness and resource security; two, tech giants investing in AI infrastructure, leading to supply-demand imbalances in fiber optics, storage, energy, and other fields. These will be key areas of attention.

********************Q10: Will the AI wave disrupt the value of research and investment?

Zhang Chuanjie: Indeed, the “information equality” brought by the tech wave will make markets more efficient. Recently, we’ve observed that market reaction times to events have shortened significantly. In this context, I believe human advantages lie in:

1. Making nonlinear hypotheses. Machines excel at processing historical data and linear extrapolation but struggle with disruptive changes. Humans can reason deeply and make extraordinary judgments.

2. Actively embracing and integrating technology, enabling human-machine collaboration to maximize efficiency.

Our firm has established an efficient research-investment linkage mechanism, with industry grouping led by team leaders to respond quickly to research needs. Communication channels are flexible, including recommendation meetings, in-depth reports, morning meetings, and other formats, fostering robust communication.