NextFin News — A pioneer of value investing in mainland China, Mr. Zhong co-founded Orient Harbor Investment with Dan Bin two decades ago, studied under commodities legend Stanley Kroll, and was among the earliest Chinese institutional investors to attend Warren Buffett’s annual shareholder meeting and meet Elon Musk. In July 2021—more than a year before ChatGPT brought generative AI into the mainstream—he was invited to deliver a keynote on AI and investing at Microsoft Research Asia’s Forum.

In an exclusive interview with Barron’s China, Mr. Zhong reflected on the evolution of value investing in the AI era, detailing what must change, what must remain constant, and how investors can navigate technological shifts without abandoning margin of safety.

“Three Constancies and Three Variables”

Q: In the AI era, what is the most critical element of value investing? Compared to the pre-AI era, what remains unchanged, and what has changed the most?

Zhong Zhaomin: I summarize the core framework as “Three Constancies and Three Variables.”

First, the three things that remain unchanged:

1. The essence of investing: Buying a stock means buying a business. Long-term enterprise value ultimately depends on a company’s ability to generate real cash flows and economic value.

2. Valuation and the margin of safety: The more transformative a technology is, the more prone the market becomes to discounting far-off narrative potential prematurely. Investors must continue to respect price; a great business is never a good buy at any price.

3. The circle of competence: AI enables individuals to acquire vast amounts of information rapidly, but expanding one's information universe does not automatically expand one's boundary of true understanding. One of the greatest risks in investing remains falling victim to unknown unknowns.

Second, the three variables that have shifted:

1. Research tools: The informational edge is shrinking rapidly, whereas the analytical edge—the gap in conceptual understanding—is widening. Previously, investment teams spent considerable effort gathering data, reviewing financial statements, and compiling documents. AI will handle these tasks with increasing efficiency. Human value will migrate toward asking the right questions, establishing cause and effect, and exercising sound judgment. AI makes data processing cheap, but it elevates the value of deep insight.

2. Economic moats: Investors historically sought static moats such as brand equity, distribution channels, proprietary tech, and high switching costs. Today, AI is rapidly restructuring these barriers. When evaluating a business, investors must ask not just how deep its moat is, but whether the speed at which it builds new moats in the AI era outpaces the decay of its traditional ones.

3. Valuation frameworks: Static modeling must evolve into dynamic, probabilistic forecasting. Because technology pathways and competitive dynamics shift continuously, investors must employ Bayesian methods, constantly updating their assumptions as new evidence emerges.

In short, while the soul of value investing remains unchanged in the AI era, its methodologies must evolve.

Bridging Coca-Cola and High-Tech

Q: Traditional value investing often relies on classic examples like Coca-Cola or luxury houses, which exhibit linear growth. How can value principles be applied to the non-linear growth of the technology sector?

Zhong Zhaomin: Many market participants narrowly define value investing as buying low-P/E, low-volatility, stable-growth companies. That is an overly restrictive view. Value investing fundamentally examines the relationship between value and price.

A beverage giant may compound value steadily over two or three decades, whereas an exceptional technology firm may generate more economic value in a few years following a technological breakthrough than it did over the preceding decade. One is linear compounding, the other non-linear acceleration, but the underlying valuation logic remains identical.

The primary task in tech investing is balancing the downside discipline of value investing with upside exposure to innovation. What constitutes downside discipline? It means refusing to pay an infinite price for unverified narratives. A tech company may operate without immediate profits or commit heavily to capital expenditures, but it must eventually demonstrate commercial validation through customers, revenue, cash flows, and return on invested capital. It must prove it creates useful products or services, generating tangible economic value rather than mere concepts.

At the same time, technology investments should not be constrained by rigid, static valuation metrics. Different levels of value investing reflect different time horizons. Short-term strategies can afford to be strict on valuation, while long-term strategies must be strict on business quality.

For the rare opportunities with immense, long-term certainty, the highest form of valuation is knowing when not to let a static spreadsheet obscure a rapidly expanding, executable long-term opportunity—such as commercial space enterprises scaling global infrastructure. True value investing is never about conservatively resisting change; it is about studying change, confronting change, and investing in change.

Navigating High Barriers in Tech

Q: Why do you view technology as an essential asset class, and how should investors overcome the sector’s steep analytical barriers?

Zhong Zhaomin: Productivity growth is a primary driver of long-term wealth creation, and technology is the single most potent force advancing productivity. AI is not merely an isolated sector; it is a general-purpose technology capable of penetrating manufacturing, software, healthcare, automotive, finance, and industrial sectors. Major industrial revolutions do not simply introduce new products; they alter how society creates value and reshape productive capacity.

However, technology investing is particularly error-prone because investors must simultaneously address three variables: whether the technology is genuine, whether the business model is economically viable, and whether a euphoric market has priced in decades of future success.

This creates a high analytical barrier. Understanding the technology alone is insufficient, as is relying strictly on historical accounting or market momentum.

To navigate this complexity, we build specialized research teams covering key technology sub-sectors—such as large language models, semiconductors, innovative pharmaceuticals, and advanced manufacturing. Domain experts conduct deep primary research, while a unified investment committee connects these insights under consistent risk-management principles.

AI further lowers the barrier for cross-disciplinary learning. Going forward, high-performing investment teams will integrate complementary technology specialists, AI-augmented analytical workflows, and structured decision-making protocols. The hardest part of technology investing is not decoding technical terminology, but translating technological capabilities into commercial metrics, and commercial metrics into sustainable investment returns.

Global Asset Allocation in the AI Era

Q: You have long emphasized global asset allocation to identify premier national assets. What unique characteristics have emerged across major global markets in the AI era?

Zhong Zhaomin: Genuine global investing is not about simply diversifying across geographic borders; it is about identifying irreplaceable, scarce assets and capabilities in specific markets.

The primary advantage of the United States remains its foundational innovation, deep capital markets, and global technological ecosystem. Consequently, key AI compute infrastructure, frontier models, and advanced semiconductor platforms are concentrated there.

China’s structural advantages lie in its engineer talent pool, advanced manufacturing capacity, integrated supply chains, and vast domestic market, excelling at rapidly scaling technologies from early deployment to mass application. Meanwhile, markets like Japan and South Korea maintain global competitiveness in specialized hardware components, advanced memory, precision equipment, and materials.

When we speak of premier national assets, we do not mean simply buying the largest enterprise in a given country, but rather targeting firms that embody that nation's core structural advantage. Furthermore, global portfolio construction should not bet on any single nation winning permanently. Global supply chains in AI are tightly interconnected: a leading software or hardware firm in the West relies on Asian memory, equipment, and manufacturing, alongside global energy and raw materials.

The central question for global investors is not predicting daily index movements, but identifying which flagship supply-chain leaders are driving the global reallocation of resources and economic value amid complex geopolitical and trade environments.

Capital Preservation and the Mechanics of Exit

Q: You recently introduced the concepts of wealth generation and wealth preservation, arguing that a complete investment framework requires both offense and defense. Why is generating returns alone insufficient?

Zhong Zhaomin: The investment industry focuses predominantly on how to make money, but a mature investment framework must also resolve how to retain accumulated gains. I refer to the former as the method of wealth generation, and the latter as the method of wealth preservation. Without effective capital preservation, an investment process lacks structural closure.

Through years of market observation, we identified two behavioral phenomena regarding execution: stop-loss orders almost always lock in an actualized loss, while profit-taking almost never sells at the exact peak.

While both actions involve exiting a position, they engage distinct psychological pressures. Managing stop-losses involves loss aversion—confronting an existing drawdown where individuals naturally resist admitting error. Managing profit-taking involves managing greed and regret—taking profits while an asset continues to appreciate requires discipline against the desire to extract further gains. Consequently, executing profit-taking is often psychologically more demanding than cutting losses.

Establishing systematic rules for stop-losses is relatively straightforward, but taking profits requires an investor to make intentional tradeoffs while gains remain on the table. Selling too early risks regret if the asset continues to rally, while selling too late risks returning capital to the market during a drawdown. This tests operational discipline and behavioral control far more than financial analysis.

This is why I place increasing emphasis on a complete investment framework comprising three pillars: theoretical foundations, macro and market understanding, and an operational system.

Acquiring theory is straightforward, and understanding the world encompasses logic across political economy, technology, and corporate governance—along with recognizing market cycles and human irrationality. The operational system ultimately dictates execution after a judgment is formed. Knowledge without a strict operational system and execution rules fails to form a complete feedback loop, leaving capital vulnerable to market reversals.

Ultimately, long-term compounding is less a test of raw intelligence than a test of operational discipline, systematic thinking, and unified execution.

(This article was originally published by Barron's China. Author: Hu Runfeng; Editor: Liu Yangxue)