Top Investor Talk | Shangya Investment Chairman Shi Bo: Don't race with quant, go to the physical world to find ten-bagger stocks, computing power is the "coal" of the AI era

Questioning AI · Shibo’s On-the-Ground Research: How Does It Reveal a Company’s True Management Quality?

Meiri Reporter: Li Na Liu Jinxu Meiri Editor: Peng Shuiping

In the spring of 2026, investors in the A-share market felt a sense of helplessness.

High-frequency trading, quantitative funds, algorithmic harvesting… Facing the invisible web woven by machines, many investors began to doubt: in this machine-dominated era, does fundamental research-based investing still matter?

Amid this confusion, Shangya Investment Chairman Shibo offered a completely different answer.

“I just came back from a research trip,” he said during a recent investor interaction, “I looked at companies making optical chips and optical modules, mainly checking their machinery, inventories, and other conditions, as well as understanding their daily management.”

While quantitative models are eager to capture price differences in 0.01 seconds, Shibo still insists on measuring industries with his own feet and observing details with his eyes. This seemingly “clumsy” approach has repeatedly helped him navigate bull and bear markets. Third-party data shows that Shangya Huoshui No.1 Fund’s returns over the past five years have far exceeded the average level of quantitative funds.

In his view, true investing is never about racing against machines but about walking with the times.

Image source: Provided by interviewee

Discussion on Investment Dimensions:

Direction Is More Important Than Effort

Over his more than thirty-year career, Shibo has formed and refined his investment philosophy, with the first being value investing.

There are many ways to achieve value investing. In his opinion, buying stocks is equivalent to buying companies. He emphasizes the importance of learning to do accounting—comparing a company’s total market value with its actual asset value to find severely undervalued companies.

But before judging whether a valuation is high or low, more importantly, one must judge the direction of the times—this itself is the first lesson of value investing.

“Chinese residents’ wealth mainly comes from real estate, from the dividends of that era,” Shibo said. “The first wave of dividends was from urbanization, from real estate. The second wave was from the internet. Currently, we are entering the AI revolution.”

In his view, the AI revolution is not a continuation of the internet but a once-in-250-years cognitive and productivity revolution. 250 years ago, Watt invented the steam engine, and humanity entered the industrial society, with physical labor amplified ten thousand times. This time, it’s the amplification of intelligence.

“The internet only solved the issue of production relations,” Shibo said. “But the AI revolution can directly generate tokens and create productivity.”

“The AI revolution is bigger and faster than the internet revolution, and it’s the most rapidly penetrating revolution in human history. It’s not the internet, nor the computer era; it’s a once-in-250-years cognitive and productivity revolution,” Shibo emphasized.

To him, the essence of investing is the realization of cognition. “Direction is more important than effort. Every wave of dividends is a dividend of the era, not of individual ability. What you need to do is identify the direction and hold on, and leave the rest to time.”

Discussion on Research:

From Workshop to Canteen, the First Focus in Research Is “People”

In Shibo’s investment philosophy, quality investing is the core principle second only to value investing. Companies that are undervalued don’t necessarily rise; the key is the quality of management. He insists on investing in companies whose management serves shareholders’ interests and possesses great corporate qualities, and on-the-ground research is precisely to examine this “people” core asset.

During his time at Huaxia Fund, Shibo was known for his “diligence.” He once researched ten listed companies in a month, condensing years of research into ten pages and 111 indicators. This consistent habit of thorough investigation drives him to continually seek the next industry turning point.

Before the gold market started, he nearly visited all domestic gold companies—from Shaanxi to Shandong—going deep into mines to see the ore. Later, he traveled to the Solomon Islands to research the global gold industry in tribal villages. This in-depth fieldwork allowed him to see the investment value of gold early. In 2023, he judged that the US dollar’s status would be impacted, and the US would enter a rate-cutting cycle, creating huge space for gold. His judgment was later confirmed as related gold stocks soared tenfold.

Even now managing private equity funds, he maintains a high frequency of research. “Now I research at least four listed companies every month,” Shibo said frankly. “And every stock I buy, I’ve been to see and investigate in person.” His words reveal his passion for research—not as a task, but as a heartfelt love.

But his research approach has its own logic.

“I must see the production line when I research,” he said. “In the compute industry chain, I look at whether the machinery is sufficient, whether equipment is running at full capacity, and how much inventory there is—these details are more real than any financial data.”

Besides the workshop, he also pays attention to often-overlooked areas. A company’s management quality cannot be seen from financial statements. Financial reports can be manipulated, but canteen conditions cannot be faked. These details can better reveal issues than profit statements. In his view, how a company treats its employees often determines how it treats shareholders.

He applies this research method to identify two types of companies.

One is the turning point companies—those at the critical juncture where supply and demand reverse, or industries moving from zero to one or from one to N. Shibo believes that identifying turning point companies relies not on financial reports but on visits along the industry chain. He checks whether upstream raw materials are sufficient, whether downstream demand has exploded, and whether competitors’ capacities are catching up.

The other is the pivot companies—those with pricing power within industries at the turning point. “When supply and demand reverse and the industry enters a turning point, you need to find pivot companies,” Shibo said. “Pivot companies are those with pricing power, with significantly higher market share or profit margins than their peers.”

In his view, differences in management quality ultimately show up in these details. Good management can seize opportunities during industry upswings and maintain bottom lines during downturns. Poor management, no matter how good the track, cannot produce outstanding results.

Discussion on the Present:

Compute Power Is the Coal of the AI Era

In Shibo’s investment methodology, scientific investing is the underlying logic throughout.

His background in investment banking gives him keen insights into industry chains, business models, and competitive landscapes. He divides industry investment into several stages:

0 to 1 is broad research. This phase resembles venture capital, pursuing high odds. The industry has huge potential, valuations expand rapidly, but performance certainty is low. 1 to 10 is in-depth research, with significantly increased certainty. At this stage, investing in pivot and leading companies is key, aiming for higher success rates. Industry barriers and first-mover advantages are critical, requiring close attention to marginal changes, especially to slowing penetration.

Based on his judgment of the AI era, Shibo has built a clear main investment line: the compute industry chain. His analysis of this line also reflects his approach of using supply-demand relationships, technological pathways, and cost curves to project industry evolution.

He believes that the US, with its technological innovation advantage, has achieved the breakthrough from 0 to 1 in AI; while China, with its strong manufacturing capacity and supply chain advantages, is poised to expand during the 1 to 10 industrialization phase. This is the core logic behind his long-term optimism for the compute industry chain.

“Compute power is the coal of the AI era, the fuel for intelligence,” he said. “The demand for compute power is growing exponentially. Every breakthrough in applications will trigger explosive demand for compute. Language models, video models, and once robots break through, the demand will be even greater. But supply is limited by the physical world—the downstream is virtual and constantly iterating; the upstream is the physical world, which can only be refined gradually. This mismatch between supply and demand is the source of investment opportunities.”

Shibo describes the value distribution along the industry chain with elasticity transmission: “Downstream chips have the highest certainty but relatively low elasticity; midstream optical modules have elasticity over five times that of chips; upstream optical chips are ten times more elastic than modules; and upstream resources like indium phosphide, tungsten, and rare earths have the greatest elasticity. The most profitable industry in the industrial age was coal, and in the AI era, the most profitable is upstream resources.”

He sees AI compute investment replacing real estate as a new engine of economic growth, representing the most important investment theme for the next 5 to 10 years.

“Global supply chains cannot do without China,” he said with an example. “Inside a compute cabinet, core components like optical modules, PCB boards, copper foil, and upstream rare metals like tungsten and indium are all dependent on China’s manufacturing capacity.”

Discussion on Strategy:

Holding On Is the Key to Excess Returns

In Shibo’s investment philosophy, the weight of “holding” is very significant. He knows that only through long-term holding can one enjoy the huge returns brought by compound interest.

During interactions, he repeatedly emphasizes the importance of holding. Most investors miss ten-bagger stocks because they lack enough research depth and cannot stay confident amid volatility. If you truly understand the research, see the right direction, you won’t panic over short-term setbacks.

But his understanding of “long-term holding” does not mean holding without moving. It’s a dynamic process of continuous tracking and validation. He sets a benchmark company for each industry he follows. If companies along the industry chain have two consecutive quarters of below-benchmark growth, he will consider selling.

This dynamic tracking also reflects his risk control system. To him, real risk management isn’t about cutting losses after a decline but about thinking through beforehand: what’s the maximum drawdown this company might face? When it drops 20%, should he reduce or increase his position? If the answer is to add, then the investment is worthwhile.

It’s worth noting that Shibo not only embraces the wave of technology but also requires his team to keep up. In daily work, he asks researchers to use “lobster” to handle emails and classify vast amounts of research reports, aiding their work. To him, this is not only an efficiency tool but also a front line for understanding industry trends. An AI-focused investment team must first use AI itself.

“Volatility is not risk; misjudgment is risk,” he said. “Risk control isn’t about stop-loss; stop-loss isn’t risk control. Risk control is about pre-judgment.”

In this era of quantitative frenzy and market confusion, Shibo believes that the meaning of investing is not to compete with machines but to walk with the times.

And perhaps this is the underlying secret that allows him to maintain clarity amid an ever-evolving era.

Meiri Economic News