36th Annual Graham & Dodd Event: A Conversation with Li Lu

• Source Video: 36th Annual Graham & Dodd Event with Li Lu ‘96, Founder and Chairman of Himalaya Capital
• Speaker: Li Lu (‘96 CC, ‘96 BUS, ‘96 LAW), Founder and Chairman of Himalaya Capital
• Moderator: Tano Santos, Robert Heilbrunn Professor of Asset Management and Finance; Faculty Director, Heilbrunn Center for Graham & Dodd Investing, Columbia Business School
• Published: October 15, 2026

Opening Remarks & Introduction

Tano Santos: Thank you all so much. I know that you are incredibly excited to be here—and I know it’s not because of me, but rather because of our guest today, whom I will introduce shortly.

Before we start with the interview, let me take a couple of minutes to share a few updates. As always at this time of the year, the Heilbrunn year is up and running in full swing. The students, some of whom are here today, have already taken my Modern Value class, and they are now taking Applied Value Investing, Credit Superhighway, and all the electives. We are having a blast, and they are all absolutely wonderful.

In addition, we have a new initiative this year at the Heilbrunn Center: we launched our first Global Value Investor Program, a joint effort with Columbia Business School Executive Education. The first cohort of that program is here with us today—wonderful friends from China who are visiting us this year. Thank you so much for being here.

Michael [Mauboussin] and I—I get asked this question a little bit too often—are already working on a new season of the Value Investing with Legends podcast. New episodes will be coming soon, so that is something to look forward to.

We also made our new stock selections for the Tom Russo Student Portfolio. This year we bought Equifax, Humana, the great Japanese bicycle component company Shimano—a classic value play if there ever was one—Copart, the online car auction company, and Perimeter Solutions. As always, you can download all the research reports the students have written over the years on these companies. It is a wonderful resource, and I very much encourage you to look at it.

I also want to single out a wonderful member of our community: I am always amazed by Tom Russo’s generosity toward the Heilbrunn Center—not only because of the student portfolio, but because of everything he does for the program, and in particular this year for making sure that our friend Li Lu could be here, for which we are deeply thankful.

We just had a wonderful week with the new Global Value Investor Program. Andrew Gundlach and Kristin Gilbertson put on a terrific session for them, which I want to acknowledge publicly. And finally, we are getting ready for a terrific CSIMA conference. The lineup of speakers coming this year is amazing: Howard Marks, Bill Ackman, Josh Wolfe of Lux Capital, and Samantha McLemore from Patient Capital will all be there. I very much hope to see many of you there.

Now to the main event of the year. Li Lu, of course, needs no introduction, as he is truly one of the great modern value investing legends. He is the founder and chairman of Himalaya Capital, which he founded in 1997, and he has built an extraordinary record of performance over the last 30 years.

As many of you know, Li Lu is one of our most distinguished alums. He graduated simultaneously with a law degree, an MBA, and a BA from Columbia College—because why get just one degree when you can get three? Apparently, that was how he thought about it. His personal story is well known, so I won’t repeat it here. Li Lu is also famous as an early investor in BYD and many other great Chinese companies, and we will have the opportunity today to discuss issues related to the Chinese stock market as well.

First of all, Li Lu, welcome, and thank you so much for being here.

Li Lu: Thank you, Tano.

Six Core Principles of Value Investing

Tano Santos: Li Lu, as I mentioned, you are one of the great names in modern value investing, and you are deeply intertwined with the story of value investing at Columbia Business School. We were talking about that over lunch—you were there at the very beginning when Bruce [Greenwald] was launching this program. We also know of your long friendship and partnership with both Charlie [Munger] and Warren [Buffett], and how many things you did together. Next year, we will be celebrating 30 years since the founding of Himalaya Capital, which is a remarkable achievement.

So I want to start here. Tell us a little bit about what you have learned about value investing, and where you think we are today. What is the thing you wish you had known at the beginning? What lesson do we have now that we didn’t have when you were studying these ideas at Columbia Business School?

Li Lu: Thank you, Tano, and thank you all for being here. It is such a wonderful occasion. Back when I was a student and a young professional, I attended a few of these luncheons myself, and they were always a highlight. I hope I can do them some justice now, 30 years later.

The discipline of value investing started right here at Columbia. When Ben Graham first started teaching these classes, it was almost 90 years ago, and he had already been practicing for 10 or 20 years before that. So we can call it a 100-year-old discipline. It has certainly evolved over time, yet some of its core principles have remained unchanged and become timeless.

If I had to distill the main principles of value investing, there are six in my mind. Three of them come directly from Ben Graham, the namesake of today’s luncheon; one each comes from Warren and Charlie; and we added one small twist of our own to make six. These six basic principles have guided our discipline and our practice over the last 33 years.

Ben Graham’s Three Foundational Ideas

Ben Graham laid out the foundation of this discipline with three basic ideas:

1. A stock is a piece of ownership, not a piece of paper to trade around. Buying a stock is the same as buying a company.
2. The market is there to serve you, not to instruct you. That is a profound insight when you are right in the midst of market activity.
3. Investing is about predicting a future that is inherently unpredictable, so you must always require a margin of safety. A margin of safety protects your downside while also positioning you for better upside.

Warren Buffett’s Circle of Competence & Charlie Munger’s Fishing Analogy

When Warren practiced value investing, he added a fourth principle: the circle of competence. All of us are probably predisposed toward certain interests, and over time you develop particular expertise, study more deeply, and refine it. The most important aspect of a circle of competence is knowing its edge—because most of us are wired to think we know a little bit more than we actually do. Understanding the boundary of your competence is just as important as understanding what lies inside it.

Charlie added a fifth idea: you have to fish where the fish are. Charlie was a lifelong fisherman, and it is a wonderful analogy for financial markets. Certain markets, industries, and businesses attract enormous attention. With attention comes intense competition, and competition averages out your returns. Meanwhile, other industries, businesses, countries, or markets simply do not attract enough attention—there are far fewer fishermen. You do way better when you fish where the fish are. It is a simple idea, but in practice it is amazingly effective.

Wealth as Percentage Ownership of Global Purchasing Power

Finally, over my own 33 years of practice as a global investor—applying value investing principles globally at a time when global commerce has increasingly become one single market—the sixth insight we arrived at is that wealth, in the end, is your percentage ownership of global purchasing power.

In a modern economy, wealth is not truly measured by the amount of gold, diamonds, or currency of any kind that you hold. With inflation, currencies dissipate over time. What truly matters is what you can buy with your wealth and what you can use it for. A far better measure of wealth is your percentage ownership of global purchasing power.

The way we express that principle is by buying great businesses in the most dynamic economies at advantageous prices, waiting long enough for value to be created over time, and sharing proportionately in that upside. Those are the six core ideas that I believe can withstand the test of time and serve every investor well.

Value Investing and the Equity Analyst in the Age of AI

Tano Santos: I like that last point very much. I am already deviating from the script you and I agreed on, because going forward, keeping an eye on purchasing power is going to be key if we are unable to control inflation in the face of today’s fiscal challenges.

Before we run out of time, I want to ask a question for the benefit of my students in the room. Value investing as a practice has seen its ups and downs, and it has improved conceptually over time under Bruce’s leadership, your leadership, and that of many others in this room. What is the future of the equity analyst in the age of AI? Do you think we can still teach what we teach at Columbia Business School going forward?

Four Types of Market Participants & The Church With a Casino Attached

Li Lu: Let me answer that in two parts: first, what investors actually do in the market, and second, how generative AI fits into the picture.

If you look at market participants, there are essentially four types of players:

1. Long-term investors, who stay with a company over time to participate in value creation.
2. Arbitrageurs, who right the wrongs of market pricing. In Ben Graham’s terms, the intelligent investor is a realist who buys from pessimists and sells to optimists.
3. Speculators, who are mostly guessing what other people will buy or sell next, without caring about either intrinsic valuation or long-term value creation.
4. Brokers and market makers, who facilitate transactions among all of the above.

In practice, most market participants are not purists; they encompass some blend of all four. An index fund is essentially the average of all of them. Value investors, however, focus on accurate price discovery and waiting for long-term value creation. I would say Ben Graham leaned more toward price-value arbitrage, whereas Warren uses arbitrage pricing at entry but relies primarily on long-term value creation.

If you look at the social utility of these four groups, it is quite revealing. Speculators, by definition, do not stay in an investment long enough for real value creation to occur. Their gains and losses sum to roughly zero; net to society, speculation is a cost to the functioning of the market, as are brokerage and market-making fees.

The primary purpose of a stock market is to channel excess savings into productive use by deserving companies, fueling the compound economic growth of a modern economy. Why, then, does the market have this massive speculative side? Because it is part of the human condition. Warren has always described the stock market as a church with a casino attached to it. The church is where value is created, while the casino is almost part of the “building permit”—a necessary cost of human nature. When stock markets were first created 400 years ago, the casino side was what inspired animal spirits and brought liquidity into the room; 400 years later, it is still doing the same thing.

This is where every student has to make a personal choice early on: do you want to play the zero-sum game, or do you want to play the value-creation game? Society faces a parallel choice: we cannot allow the casino side to grow so out of proportion that it overwhelms the primary purpose for which the market exists. The fact that over 400 years—despite massive booms, busts, and waste—this mechanism has survived and continued to propel human society forward is nothing short of miraculous. The two sides coexist as part of the human condition, but each of us must choose which side we want to stand on.

Why Generative AI Is Different: The Spectrum of Tool vs. Creature

Now let’s turn to generative AI and the market. First, what is this wave of generative AI?

Ever since the Industrial Revolution a few hundred years ago, humanity has experienced wave after wave of technological disruption and creative destruction. Every technological wave is different, yet this wave feels unusually different from anything we have experienced before.

On the surface, the pace of change, the scale of investment, and the extreme concentration of financial impact at the very top are all unprecedented. Why is that? Because the underlying narrative is fundamentally different.

Until now, almost every technology in human history has been understood as a tool to enhance human productivity. With generative AI, a new dimension has entered the picture: a spectrum between tool and creature. And we are beginning to slide along that spectrum.

When large language models burst onto the scene nearly four years ago, they effectively passed the Turing test: we could no longer tell from language alone whether we were interacting with a machine or a human. Next came reasoning, giving the technology increasingly human-like cognitive features. From reasoning came coding, and with coding capability, AI became an autonomously executing agent that functions much like a human worker. Now we are seeing multi-agent collaboration across complex, long-horizon tasks—what is called long reasoning—where agents organize, self-organize, prioritize, and execute tasks using their own communication protocols, much like a human society.

According to the prevailing Silicon Valley narrative, generative AI is now entering the stage of RSI (Recursive Self-Improvement), where future improvements are generated by AI itself without a human in the loop. From RSI, the narrative goes, we will move to AGI (Artificial General Intelligence) and then ASI (Artificial Superintelligence)—intelligence exceeding that of all humans, both individually and collectively. At that point, we would effectively be creating a man-made demigod. Because such a creation would be all-encompassing and omnipotent, its economic impact would by definition be infinite, justifying any amount of capital expenditure today. So when you see trillion-dollar valuations and upcoming IPOs, you are practically being invited to participate in the “IPO of a god.”

That is the narrative driving the market. But step back for a moment and ask: if that narrative were actually true, what would become of humanity? We would be living in a society lorded over by a man-made god—accessed by a tiny few, understood by none, and controlled by no one (since you cannot control what nobody understands). Would any human being actually want to live in that society? The answer is a resounding no.

Therefore, there are really only two possible outcomes:

• Outcome A: This technology is not actually progressing toward godlike superintelligence—and we can examine why.
• Outcome B: If it does progress in that direction, humanity will inevitably unite at some point to stop it from becoming a sovereign creature over us.

In neither outcome will human society end up being ruled by a man-made god.

Provisional Definition of Generative AI & The Biological Gap

Leaving the “demigod” narrative aside, is generative AI a genuinely useful technology? The answer is a resounding yes.

If I were to give a provisional definition based on observing where it came from and what it does: generative AI is a machine that applies brute-force compute and clever algorithms to all human-generated data (and its first derivative) to establish systematic, contextual correlations across near-infinite variables. In doing so, it replicates a substantial portion of human reasoning and intelligence. At present, we know how it works and how to scale it, even if we don’t fully understand why it works—much as we still don’t fully understand how the human brain works as biological machinery.

Viewed through that lens, generative AI is a powerful synthesizer and equalizer for a specific layer of the economy. Think of the economy as a stack of layers:

• At the foundation are energy and transportation.
• On top of that sits manufacturing—combining atoms and molecules into physical goods.
• Above that sits communications and digital media.
• At the top sits a vast layer of knowledge-based and technology-based services, which synthesize accumulated human knowledge to serve all the other layers.

At its current stage, generative AI is an extraordinary tool for synthesizing and equalizing the knowledge layer. But it does not directly alter the physics of energy, transportation, or atomic manufacturing, and digital communications are already highly synthesized. Moreover, because it operates strictly within the realm of human-generated data and its first derivative, it has not transcended human bounds.

To reach the complexity of biological life requires a completely different order of magnitude. A single human cell contains roughly 50 gigabytes of information, and you have to multiply that by trillions of cells in the human body. Current AI models are nowhere near that biological scale.

The Human-in-the-Loop Red Line & Incremental Cost

Even so, what AI has already achieved is beginning to tip the balance between tool and creature. Recursive self-improvement (RSI) without a human in the loop is extraordinarily dangerous: left to define its own objectives, an autonomous system will generate goals and sub-goals aligned with its own purposes, drifting further and further from human control.

In summary, this technology is at once enormously helpful, transformative, disruptive, and extraordinarily dangerous. As a human society, we must draw a firm red line: no self-learning or autonomous tool utilization without a human in the loop. AI must remain firmly on the tool side of the spectrum, never the creature side.

Beyond existential safety, AI brings massive societal disruptions—from employment displacement and emotional dependency to cyber, biological, and nuclear risks. Society will need a balanced, measured pace of adoption that weighs benefits against risks.

There is also a crucial economic distinction: unlike the internet or prior software waves—where a one-time upfront investment could be amortized over many years at near-zero marginal cost—generative AI carries permanent, substantial incremental costs (compute, energy, chips) for every unit of output. Because of this ongoing incremental cost, capital needs, revenue, profit, and financial discipline will always remain in force. Between financial discipline, physical bottlenecks, and societal tolerance, human society won’t adapt through smooth central planning; we will go through major boom-and-bust cycles—perhaps several of them—as this wave plays out.

Tano Santos: What you just said about AI being both extraordinary and dangerous—and our not fully understanding why it generates some of the outputs it does—implies that we will always need a skilled human in the loop to verify its output. That means the scarce skill of the person capable of monitoring AI and communicating effectively with it inside a specific domain will determine where economic rents accrue. Do I understand you correctly?

Li Lu: Yes, that is very much along the right lines. There are different nuances across industries, of course, but in the end, domain judgment with a human in the loop is critical.

Market Concentration, Irrationality & Long-Term Mean Reversion

Tano Santos: Let’s project these ideas onto the financial markets. Many of my colleagues here teach capital markets and investing, and when we walk into the classroom today, we are looking at a remarkably peculiar market.

The market is being driven by a tiny handful of companies pushing the technological frontier, and valuations have been heavily distorted as a result. There is a chart I show my students tracking the market capitalization of the largest company in the world normalized by US GDP. Historically, that ratio has stayed between 2% and 4%; today, with Nvidia, it is around 15%. How do you think about a market that has become so dependent on so few companies, and how do we design portfolios to protect ourselves against technological shocks?

The “IPO of a Demigod” Prospectus & S&P 500 Long-Term Returns

Li Lu: Markets have always done peculiar things, and they always will. One of Ben Graham’s greatest contributions is the insight that you do not have to be dictated by what the market decides to do at any given moment.

Consider a fundraising prospectus I recently read from a leading AI company. On the one hand, the company calls itself a research lab, has burned through tens of billions of dollars, and promises to burn through hundreds of billions more to build a transformative technology. On the other hand, in the very same prospectus, it states that this technology carries a 10% chance of literally killing everyone—and then invites investors to put in at least $100 billion at a$2 trillion valuation!

Think about the irony and internal contradiction of that proposition: those two claims cannot simultaneously be true as an investment thesis. It is an extreme illustration of how irrational the market can become when gripped by a narrative. And history tells us that what is irrational and unsustainable eventually stops.

Over long horizons, market returns are anchored by economic reality. Take the S&P 500: over the 16 years since 2010, it has compounded at roughly 14% to 15% per year. Yet in the preceding 10 years, from 2000 to 2010, its return was zero—in fact, slightly negative. If you average the two periods across the full 26 years, you get a high-single-digit annual return.

That long-term average makes complete sense: if real GDP grows at roughly 2% and inflation runs at 2% to 3%, nominal GDP grows at 4% to 5%. Publicly listed companies—especially the 500 best companies selected through competitive elimination—do somewhat better than nominal GDP, bringing long-term equity returns to the high single digits.

Market Bifurcation and the Opportunity for Rational Investors

That high-single-digit anchor is the fundamental gravity of the market: over the long run, investors get roughly what the underlying economy produces. Periodically, however—and sometimes for extended stretches—the market swings to extremes.

If society ultimately refuses to tolerate a “super-creature” and keeps AI within the realm of tools, then AI will be evaluated like any other tool—on its costs and benefits—and capital market discipline will reassert itself through multiple boom-and-bust cycles.

In the meantime, extreme concentration creates extreme bifurcation. At any given point in time, the total pool of capital is finite. When capital is overwhelmingly sucked into a handful of mega-cap names at the top, a vast number of otherwise wonderful businesses are left trading at very low valuations. We see this bifurcation in the US, in China—where it is perhaps even more extreme—and around the world.

For an investor who can stay genuinely rational, this bifurcation actually creates a much richer opportunity set than usual. Of course, staying rational during a massive technological wave is hard: when the margin of imagination expands dramatically, speculation runs wild and exaggerations become extreme. Still, over the long haul, the natural weight of market rationality always prevails.

China’s Economy: The Middle-Income Trap, Industrial Upgrading & “Objective vs. Cost”

Tano Santos: You mentioned China briefly, so let’s talk about that. I just came back from teaching our program participants in Hong Kong, and I used the occasion to write a couple of case studies on Chinese companies and study the Chinese market. It has been quite a ride over the last few years. With the real estate bust and strains in the banking system, parts of it look reminiscent of balance-sheet recessions we have seen elsewhere. How do you view the Chinese economy and investing in China today?

Breaking Through the Middle-Income Trap & Made in China 2025

Li Lu: If you look at the Chinese stock market over the last 10 years, it has largely stagnated. Yet what has happened underneath the surface during those 10 years is fascinating.

Ten years ago, China had reached a middle-income level and faced a defining historical question: how do we break out of the middle-income trap? Dozens of countries around the world have industrialized into low- to mid-end manufacturing, hit the middle-income ceiling, and remained stuck there for 40 or 50 years—or longer. Left to its natural course, a developing economy rarely breaks through that ceiling on its own.

To avoid that fate, the Chinese government made a deliberate decision to migrate the economy from low- and mid-end manufacturing into mid- and high-end manufacturing. Instead of a standard five-year plan, they launched a 10-year blueprint—Made in China 2025—targeting roughly 10 strategic industries for foundational upgrading.

Just a few years after that plan was launched, the first Trump administration took office and initiated the US-China trade war. To Beijing, that reinforced the conviction that not only is breaking through the middle-income ceiling difficult on its own, but incumbent powers might actively try to hold China beneath that ceiling. The Biden administration subsequently kept the tariffs and added sweeping technology restrictions.

Consequently, five years into the 10-year plan—around 2021—the Chinese government doubled down. On top of upgrading manufacturing to the mid-to-high end, they elevated technology self-sufficiency to a top national priority across every major bottleneck: semiconductors, large aircraft engines, foundational operating software, and more.

The Real Estate Adjustment & Household Balance Sheets

Pursuing goals that almost no other developing country has achieved comes at a deep and heavy cost. In fact, that has been China’s development model for the past half-century: every decade or so, the state sets a paramount strategic objective, and another part of the economy bears the cost.

Over the last five years, the principal cost of financing China’s industrial and technological push was borne by middle-class household balance sheets through the real estate adjustment. When Chinese property prices peaked in 2021, a typical middle-class Chinese family had roughly 70% of its net worth tied up in housing, backed by roughly 50% mortgage leverage. Over the subsequent five years, nationwide property prices fell by roughly 40% on average—meaning a typical leveraged middle-class family saw 60% to 70% of its net worth evaporate.

When a family experiences a balance-sheet shock of that magnitude, it does three things: consumes less, saves more, and pays down debt wherever possible. That is exactly what typical middle-class Chinese families have done over the past five years. As a result, domestic consumption growth slowed steadily toward zero, while household savings surged to roughly 165 trillion RMB—around 120% of GDP, the largest savings pool in the world. Combined with two to three years of flat to slightly negative price levels, consumer confidence has been understandably subdued. That is one tier of China’s two-tiered economy.

Semiconductor Sanctions, AI Code Generation & The Two-Tiered Economy

If you look at the other tier of the economy—the strategic objectives pursued over the last five to 10 years—you see a remarkably different picture. Chinese manufacturing has decisively broken through from the low-to-mid end into the mid-to-high end. Every single major target of Made in China 2025 has been met.

In the last five years, China has also made dramatic headway in resolving its technology bottlenecks. Take semiconductors: ironically, US-led technology sanctions turned out to be a godsend for China’s domestic semiconductor industry, because sanctions handed domestic chipmakers the single most precious thing in a market economy—guaranteed customer demand. Ask any entrepreneur what matters most, and they will tell you it is the order book. Once domestic orders flowed in, the flywheel started turning: revenue generated profits, profits funded R&D, R&D drove rapid iteration, and better products brought even more revenue. Today, in mature process nodes, China has achieved near-complete self-sufficiency across both semiconductor equipment and fabrication; in advanced nodes, it is closing the gap rapidly with only a few missing links remaining.

Generative AI has acted as another unexpected equalizer. One of AI’s strongest capabilities is automated code generation at scale, which directly addresses one of China’s hardest historical bottlenecks: foundational systems software. From PC and server operating systems to alternatives to Nvidia’s CUDA software stack, AI-assisted coding is accelerating domestic replacements across the board.

Across field after field—electric vehicles, batteries, energy storage, renewable energy equipment, heavy machinery—Chinese companies have become dominant global leaders. And the macro numbers tell the story of the resource transfer: annual new home sales in China are down roughly 70% from the peak—a drop of roughly 10 trillion RMB per year—and that exact sum, roughly 10 trillion RMB a year, has been redirected into manufacturing upgrading and technology self-sufficiency since the 2021 inflection point.

China’s 10-Year Cycle: Are You the Objective or the Cost?

If you step back and look at the last 50 years of Chinese development, this pattern repeats roughly every 10 years: the country sets a new strategic objective, achieves it at a heavy cost, and in the following decade, yesterday’s cost often becomes tomorrow’s objective—while yesterday’s objective becomes tomorrow’s cost:

• During the early industrial takeoff, the environment—air, rivers, soil—was the cost. A decade or two later, environmental restoration became the state objective, spawning world-leading green technologies, while polluters became the cost.
• During the decades of rapid urbanization, real estate developers were the objective, building modern cities, high homeownership, and middle-class wealth. Over the last five to 10 years, real estate became the cost.

That mechanism is how China compressed several hundred years of Western industrialization into roughly 40 years. For any entrepreneur or investor operating in China, the most important question at any given time is: are you the objective, or are you the cost? Being on the “cost” side of a state objective is no fun at all.

Overall, China has now emerged with the scale, industrial depth, and talent base to embark on full modernization. Yet because of the two-tiered economy and bruised household confidence, its stock market has gone nowhere for 10 years—leaving some of the world’s most globally competitive companies trading at rock-bottom valuations. For a rational investor who looks at the whole picture, today’s bifurcated Chinese market offers better long-term opportunities than you would find in more euphoric times.

Changing World Order vs. The Global Common Market

Tano Santos: While the audience gets their questions ready, I want to ask you one last question. You founded Himalaya Capital in 1997, right around the time I arrived at the University of Chicago as a young assistant professor. Back then, the world felt very different from how it looks today: it was open, transnational, and had a certain generosity of spirit that seems lacking now. How do you see the global order evolving going forward, and how does that shape the way you think about investing?

Political Order vs. The Inescapable Economic Common Market

Li Lu: That is a great question, Tano. Over the last 30 years, the world has unquestionably changed. Political world orders—how the world governs and organizes itself—shift on a slow, multi-decade or century-long scale. The post-World War II order emerged from the near-collapse of civilization across two world wars, evolved into an American-led order, consolidated into a unipolar order after the Cold War, and is now transitioning once again into a new configuration whose final shape has yet to emerge.

Beneath the political order, however, another order has been steadily forming over the past several centuries: the economic order. An economic order forms when more and more countries and populations engage in common trade and business transactions, creating a common market. With the collapse of the Soviet Union and the end of the Cold War roughly 30 years ago, virtually the entire world population joined a single global market economy for the first time in history.

What is remarkable about a global common market is that once formed, it is exceedingly difficult for any single country to withdraw from it voluntarily. Participation in the common market drives an unstoppable flywheel of technology, division of labor, resource allocation, and efficiency. Any nation that isolates itself from that common market for an extended period pays a crippling price.

Look at a century of trade wars and tariffs: no matter what barriers governments erect, commerce finds a way to flow through third countries along the lines of comparative advantage. Headline bilateral trade between the US and China has declined, for instance, yet if you look at US trade with Mexico and ASEAN, it has risen by almost the exact same amount. The underlying trade hasn’t disappeared; it is simply routing through third-party countries.

In the political realm, two major shifts are underway:

1. First, the United States—having led the global order for 80 years and shouldered most of the cost of policing the global commons—has begun to feel that the costs outweigh the benefits, and is attempting (not very smoothly so far) to step back from some of those responsibilities, leaving a geopolitical vacuum.
2. Second, the single biggest structural change of the past 40 years is the rise of China as a massive economic entity with a full-spectrum manufacturing and technology ecosystem and the world’s largest unified, regulated domestic market.

The Semiconductor “Chopsticks” Reality & Strategic Stability

Yet even as the political order restructures, our mutual economic dependency within the single global market has only deepened.

Take generative AI—the technology on which the entire US market is currently betting. The single most indispensable physical input for AI is semiconductor chips. Think about every country and region on earth that fabricates semiconductor chips. They all share one thing in common. What is it?

Tano Santos: You’re putting me on the spot! Now I have to think about semiconductors—what do all semiconductor producers have in common?

Li Lu: Every single place that fabricates the semiconductor chips essential to this technological wave—and all the people who make them—use chopsticks. Think about that: without exception, in fabrication. And for good measure, the CEOs leading the major semiconductor companies in the United States—from Jensen [Huang] to Lip-Bu [Tan], Lisa [Su], and Hock [Tan]—all use chopsticks too!

That is a vivid illustration of mutual dependency. If a country and its companies are betting everything on this technological wave, whatever its promise and perils, you have to be nice to people who use chopsticks. That reality did not happen by accident, and it is not changing anytime soon.

There is a reason the President of the United States and the President of China have felt an urgent need to meet four times within an eight-month span—where in two of those meetings, the single most important deliverable was simply scheduling the next meeting.

This brings us to the concept of strategic stability. Peace and stability are of paramount strategic importance to every major power because all of them feel the weight of the single global common market and the prohibitive cost of withdrawing from it. Meanwhile, of course, everyone is competing fiercely, each believing time is on their side—China racing to complete semiconductor and technological self-sufficiency, the US betting on frontier AI breakthroughs, and other nations pursuing their own calculations. Yet beneath that competition, the largest nations share both the responsibility and the self-interest to preserve strategic stability. The post-WWII political order is transforming, but the global single common market is here to stay.

Building Value Investing Cathedrals Globally

What does this mean for us as investors? It brings us right back to our sixth principle of value investing: as a global investor, your wealth is measured by your percentage ownership of global purchasing power. And the best way we know to preserve and grow that share is by buying great businesses in the most dynamic economies at advantageous prices and holding them for the long term.

To return to Warren’s metaphor of the stock market as a church with a casino attached: what Berkshire Hathaway has accomplished over the last 60 years is to build that church into a gigantic cathedral—a place that not only exemplifies how capital allocation should function, but also offers spiritual solace to investors who believe in what the market exists to do.

What we at Himalaya have tried to practice over the last 30 years is to prove that this gospel can be spread globally—that we can build value investing churches in Asia and other regions, some of which may grow into cathedrals in their own right over time.

And that brings me back to your question about the students and young professionals in this room. In a market structured as a church with a casino attached, the choice is yours:

• On one side lies the win-win approach of long-term value creation, which delivers a net benefit to society.
• On the other side lies the zero-sum game of speculation, which imposes a net cost on society.

Technology, including AI, will remain a tool rather than a replacement for humanity. If it ever threatens to replace humans or become a man-made god, humanity will unite to stop it—and as investors, you certainly don’t have to participate in a prospectus pitching the “IPO of a demigod.” Upholding rational capital allocation is one way we fulfill our collective duty to a good society, and that is why the Heilbrunn Center’s program exists and ought to continue. Thank you very much.

Audience Q&A

Round 1: Chinese vs. US AI, Stock Market as Objective or Cost & Political Risk

Ed Lugo (EL2 Capital): Thank you again for sharing your thoughts. Could you give us your sense of the core difference between Chinese AI and American AI?

Wenshan (Johns Hopkins): Following up on your framework for China’s economy: do you think the Chinese stock market will become the “objective” or the “cost” going forward?

Dan Bakalarz (Unison): Given your long track record of investing successfully in Chinese businesses, how do you underwrite political risk when regulatory rules can change overnight?

Li Lu: Those three questions—China’s approach to AI, whether the Chinese stock market is an objective or a cost, and how to underwrite political risk—are all closely connected.

1. Silicon Valley AI vs. Chinese AI

If you look at the Silicon Valley approach to AI, the fundamental driving question is: “What else can this technology do?” That is why the leading US players still call themselves research labs and talk about the “frontier”—they want to push the outer boundary of capability as far as possible.

Chinese AI companies are curious about frontier capabilities too, but they ask a second, even more central question: “What can this technology do for my customers—both consumers and enterprise clients—right now?” That practical, application-first orientation leads to a very different development path and very different operating metrics.

2. Is the Chinese Stock Market the Objective or the Cost?

At different times, policy emphasis shifts. But think about the balance sheet of both Chinese households and the Chinese nation as a whole: as real estate’s share of household wealth declines from its historical 70% peak—not that real estate will fall forever, since housing has real utility value, but its proportion is permanently coming down—what asset class can take its place to store and grow the wealth of families and the nation?

When you ask that question, the answer is clear: sooner or later, it has to be the stock market. Equities are the only major asset class that grows with the economy, grows because of the economy, and in turn helps drive economic growth in a mutually reinforcing flywheel.

3. Underwriting Political Risk

Understanding political risk comes down to studying a country’s history, examining its track record, and accurately identifying its core national objectives at any given stage. And that discipline doesn’t just apply to China—it applies to the United States and every other country we invest in. Our world is still organized around sovereign nation-states, and every investor must apply that analytical lens wherever they operate.

Round 2: Growth & Tech Investing, Energy & Climate, Crypto & Career Risk

Daniel Qi (New Jersey): Can value investing principles and strategies be applied to investing in high-growth or high-tech companies?

Audience Member: Energy markets are increasingly correlated with the compute consumption of knowledge-based AI businesses. How do you view the convergence of energy and intellectual capital, and its implications for climate and environmental goals? Also, what are your thoughts on cryptocurrency markets in the context of global finance?

Haoming Li (ACT Capital, Columbia Alumnus): I have been practicing in the “church of value investing” for two years now. How do you think about the risk of underperformance—when you preach value, watch speculative assets race ahead, and face very real career and client-redemption risks?

Li Lu: Let me take all of those in turn.

1. Value Investing in Growth and Technology Companies

I believe all intelligent investing is value investing, because in every intelligent investment you are trying to receive more in value than you pay in price. Price is what you pay; value is what you get. Whether you are buying a stagnant or declining business that returns massive cash relative to its price, a rapidly growing technology company, or even a portfolio of asymmetric venture bets, the underlying math is identical: the discounted present value of future cash flows over the life of the business, to the extent you can understand and estimate them accurately.

So the fundamental principle never changes. What determines whether you can invest in a high-tech or high-growth company is Warren’s concept of the circle of competence—whether your specific circle of competence allows you to estimate those future cash flows with confidence.

2. Energy, AI Constraints, and Climate

Energy is the foundational layer of the entire economy—beneath transportation, manufacturing, communications, and knowledge services. Right now, the binding constraints on AI scaling are physical: chips, electricity, and skilled labor. How energy supply evolves to meet AI compute demand will have profound implications for every layer of the economy as well as for climate targets.

3. Cryptocurrencies

As for crypto, I have a very simple answer: I don’t think anything of it at all. I think it is fundamentally bogus and will eventually disappear on its own.

Ask yourself what makes a currency redeemable and trustworthy as a medium of settlement and store of value: it is backed by the sovereign force and legal enforcement of a state. Without sovereign enforcement, a digital token conjured up out of a garage cannot function as a true currency or reliable store of wealth. Thus far, its primary real-world utility has been facilitating illicit transactions and criminal activity; society would frankly be better off banning it. If individuals want to collect crypto tokens the way art collectors bid up a particular artist’s paintings, that is their personal preference—but in terms of social utility, I think nothing of it.

4. Career Risk, Underperformance, and the “Mutant Gene”

Your question about career risk is very real. Even when you personally buy into the philosophy of value investing, you still need your clients and institutional partners to cooperate with you.

This goes right back to human psychology: everybody wants the best of both worlds. People say, “I want to practice long-term, win-win value investing—but I also want to report strong relative numbers every year, preferably every quarter, every month, every day, and every hour.” If you try to satisfy that demand, pretty soon you become a speculator.

You cannot have both worlds. Either you believe the market is there to instruct you, or you believe the market is there to serve you—you can only choose one. Anyone who tries to mush the two together would be much better off just buying an index fund, which is literally the average of all four types of market participants.

To build an enduring investment business around true value investing, two things are required:

1. First, you have to be a natural contrarian—you almost need a “mutant gene” where your conviction comes strictly from logic and facts, completely unaffected by whether the crowd agrees or disagrees with you. If you don’t have that psychological wiring, don’t force it.
2. Second, you must carefully select like-minded capital partners who genuinely share that time horizon. That means you cannot set out saying, “I want to raise the biggest fund and maximize AUM.” It doesn’t work that way. You attract the partners you deserve, and over time you get the results you deserve.

Round 3: Compounding Knowledge, Healthspan, Deserved Trust & Margin of Safety

Alex Captain (Cat Rock Capital): After such a long and successful investing career, what have you learned over the last 10 years that makes you a better investor today than you were 10 years ago? And if I can sneak in a second question: at 60 years old, you keep your mind and body remarkably sharp—taking three questions at a time and remembering every nuance. What are your daily or weekly routines?

Audience Member: Looking ahead a decade or more, what are the most important structural changes we can expect in the US and Chinese capital markets?

Wyatt (First Manhattan): When you are underwriting an investment, how do you ensure that you have an adequate margin of safety and downside protection?

Li Lu: From the mundane routines of daily life at 60—thank you for reminding me of my age!—to margin of safety and long-term structural changes, let me weave these together.

1. Margin of Safety and Expanding the Circle of Competence

One of the defining characteristics that separates a value investor from most other market participants is the enormous attention paid to downside protection. Because the future is inherently unpredictable, you protect your downside by investing only in what you truly know.

When we are young, what we truly know is very limited. When I started investing in my 20s, I didn’t know much at all, so I stuck to what was visibly there on the balance sheet—just as Ben Graham taught—because balance sheets are much easier to verify than distant growth projections. Over time, I progressed gradually from simple situations to more complex businesses.

There are two sides to the circle of competence: first, rigorously respecting its boundary at any given moment; and second, consciously and steadily expanding that boundary over time. I have always believed that if you can compound your knowledge in a sustainable fashion, eventually you will compound your capital as well.

2. Compounding Knowledge, Healthspan, and the “Seamless Web of Deserved Trust”

That directly answers what I have learned over the last 10 years. Every 10 years, I find that I have learned so much that I am almost a different person compared to a decade earlier. That sense of continuous renewal is one of the greatest rewards of choosing a win-win path over a zero-sum game.

In a zero-sum game, human nature pulls you into the ultra-short term: whatever you learn today becomes obsolete tomorrow. Nothing accumulates. You rely increasingly on impersonal tools—faster servers, bigger computers, and now AI—until eventually you yourself are replaced by a machine. What is the human value in that?

When you play the long-term, win-win game of value creation, your knowledge compounds, your judgment compounds, your relationships compound, and—while it may not extend your biological lifespan—it certainly extends your productive span and healthspan, because your mind stays deeply engaged in work that is fascinating and genuinely useful to society.

Most importantly, you compound trust with the people around you. My mentor, partner, and teacher Charlie Munger always advocated seeking the highest order of human relationship: what he called “a seamless web of deserved trust.” Approaching life and business through a seamless web of deserved trust is a complete philosophy for living a productive, win-win life. It may look slow in the beginning and take years to build, but slowly and surely, it takes you exactly where you deserve to go.

3. Luck, Compounding, and Never Losing What You Already Have

Of course, life and investment records are always shaped by a mix of lucky and unlucky breaks. I have never believed that our track record was purely the product of our own genius; we benefited from many lucky breaks along the way.

Yet if you choose the right partners, the right philosophy, the right people to back, and the right things to do, the power of compounding over decades is extraordinary. And the most profound rule of compounding is simple: you don’t lose what you already have.

Think about that: don’t lose what you already have. That applies not only to capital, but to your knowledge, your trusting relationships, your wisdom, and your health.

In that fundamental sense, the underlying structure of the US and China is the same—because it is the structure of humanity itself. Change is the one constant in life, and every era brings its own share of good luck, bad luck, and great opportunities. Finding something you naturally lean toward, and building on it day after day in a cumulative, compounding fashion alongside people you trust, is simply the best way to live a productive life.

Tano Santos: Thank you. Thank you so much, Li Lu. This has been wonderful.