But yesterday, the reality set in. Writing has aggravated a pain issue I’ve neglected for too long. Before I write half-heartedly, or this newsletter becomes something less than what it should be, I’m stopping.

I’m ending The Great Wall Street.

It may sound like an impulsive decision. It isn’t. Understanding does not grow linearly; it comes with singularities. Mikhail Gromov put it best:“Either you have no inkling of an idea, or, once you have understood it,the very idea appears so embarrassingly obvious that you feel reluctant to say it aloud.”

I understood.

I still love this work. It’s more interesting to me than ever. I just won’t be putting it down in written form anymore.

To everyone who read, subscribed, or reached out—thank you. I hope those conversations continue in whatever form they find.

If you’re a paid subscriber, you’ll receive a pro-rata refund automatically within the next few days. No action needed on your end.

Happy New Year.

Read more

And yes, one thing seems obvious: AI will be massively transformative, and the use cases will go well beyond what any of us can list today. But that statement is exactly what people said about the internet in 2000, and it was correct then too. The internet was transformative and also in a bubble at the same time.

So the better question is not whether AI matters. It does. The better question is where to position yourself within the inevitable volatility.

Some companies say they cannot build fast enough. Researchers warn about a bullwhip effect across different timescales: capital gets committed quickly, GPU orders pile up quickly, hype spreads instantly, but electricity, grid connections, transformers, permits, and data centers do not move on internet time. When the system has mismatched clocks, you do not get a smooth S-curve. You get volatility and surprise. Short sellers point out the discrepancy between Capex and earnings.

A Simple Example of Embedded Volatility: NVIDIA

Look at NVIDIA as a mental model for how volatility can be structural rather than accidental. The story sounds great until you look at who is paying, and why. If you can sell silicon at 75 percent gross margin, the obvious question is whether that margin is actually durable when a meaningful share of demand is coming from customers that are, to put it politely, not exactly printing money.

If one of the biggest customers is a loss-making company that has announced $1.4 trillion of investment in data centers (with GPUs as the biggest cost factor) while having virtually no revenue relative to that $1.4 trillion and no proven business model, then the supply chain is not anchored on stable end-demand. It is anchored on funding cycles, narrative cycles, and competitive panic.

And there is substitution. The Googles, Alibabas and Amazons of the world do not enjoy paying 75 percent gross margins to anyone. They build their own chips, they optimize around their own workloads, and they use their distribution to make their economics work.

None of this says NVIDIA is a bad company. It says the ecosystem can swing from shortage to digestion faster than investors want to admit, especially when power and deployment bottlenecks can delay real consumption even if the orders were placed.

Three Ways to Invest in a Technology Wave

Historically, every major technology upgrade gives you three broad ways to invest.

One is to build the core infrastructure itself: the CapEx-heavy backbone, the fiber cable, the railroad. In AI terms, that is the model layer and the training-and-inference buildout that burns capital to create capability.

The second is the pick-and-shovel trade: you sell the tools, inputs, and enabling infrastructure to the builders, like chips, cooling, power equipment, and data-center construction.

The third is the application layer: once built out you use the infrastructure cheaply to build products and business models that were not feasible before.

The pattern is not guaranteed, but it has a rhyme. The capex-heavy builders often struggle because returns get competed away and cycles are brutal. The pick-and-shovel players often win early because everyone rushes to build at once, then growth normalizes when the buildout saturates. Once the technology is ready the application layer tends to capture the durable value because it turns cheap infrastructure into new profit pools.

Caveat: AI might differ in one important way

Read more

This is a brief update on a company I’ve written about before.

Read more

One of Peter Lynch’s most quoted lines is:

Selling your winners and holding your losers is like cutting the flowers and watering the weeds.

Warren Buffett endorsed it, and in finance, a Buffett endorsement turns advice into scripture. The principle became eternal wisdom, repeated until the end of time.

The logic is sound: let your winners compound, cut your losers short, never throw good money after bad. It is the foundation of disciplined investing, and I follow it—in most places.

But the market’s deepest irony is that it eventually punishes every rule followed with religious devotion. If genius could be bottled in short quotes, we’d all be rich and bored. The real work lies in knowing when the rule no longer applies.

This became clear in my recent year-end review - yes I do the review beginning of December. A very simple strategy in my Chinese holdings has boosted performance for the last three years. When I mention it to other investors, they push back. It sounds wrong.

So I’m writing it down—if only as food for thought or to document my folly for the future.

Subscribe now

The Inversion

In my Chinese holdings, I water the weeds and cut the flowers. I do not do this elsewhere. China is different.

The mechanism is simple:

1. I establish a core position at a price that reflects a long-term valuation thesis.

2. If the price falls significantly (say, 25%) without a deterioration in the thesis, I add meaningfully.

3. Upon a full recovery to my original entry price, I sell the added shares, banking the profit and reverting to my core position.

4. I apply the same logic in reverse: trimming on violent, sentiment-driven spikes that dislocate from fundamentals.

I am not trading the company. I am trading against the emotional volatility of its shareholder base.

Why This Has Worked in China

This approach has helped over the past few years because Chinese stocks, particularly ADS and H-shares, have been volatile and range-bound. They swing violently, often disconnected from business fundamentals.

Just holding through these swings means enduring panic and euphoria for nothing. But by adding during panic and trimming during euphoria, you harvest a premium from the chaos. You get paid for being the calm counterparty to someone losing their mind.

Share

The Foreign Holder Problem

I believe in large part this pattern exists because of who owns these shares. They’re held primarily by foreigners—investors who can’t read Chinese, consume the country through Western headlines, and have never used the products or walked the streets. The result is predictable: overreaction in both directions.

A variant of this plays out with Ping An. I maintain a core position in the A-shares. When a negative headline sparks a panic sell-off in the H-shares, the discount between them widens to absurd levels. I then sell the A-shares and buy the structurally identical, but cheaper, H-shares. When the panic subsides and the gap closes, I reverse the trade. Same principle: get paid for staying calm while others panic.

The Crucial Caveat: Where the Rule Still Reigns

This is niche advice. In a market characterized by strong, persistent upward trends—like the U.S. over the last decade—this is a recipe for ruin. You would prematurely sell your Amazon and average down into a value trap. In a bull market, Lynch’s wisdom is not just correct; it is the only wisdom that matters.

My point is narrow: in a volatile, sentiment-driven, and range-bound market, the inverted rule can work. For Chinese offshore stocks, that has been the reality.

A Practice, Not a Principle

In investing, everything works until it doesn’t. The edge you discover persists only until the world changes. The market sends no memo.

How long will I keep doing this? I do not know. I am certain, however, that I will eventually have to stop. The day may come when I must relearn to let my Chinese winners run.

A Final, Unlikely Bit of Wisdom

A friend recently pointed me to a speech Taylor Swift gave at an NYU graduation. (I wouldn’t know a Taylor Swift song if it hit me in the head, but I know she’s dating Travis Kelce of the Kansas City Chiefs. That’s the full extent of my expertise.)

One passage, however, applies remarkably well to the fundamental dilemma of investing:

“Every choice you make leads to the next choice, which leads to the next, and I know it’s hard to know which path to take. There will be times in life where you need to stand up for yourself, times when the right thing is actually to back down and apologize, times when the right thing is to fight, times when the right thing is to turn and run, times to hold on with all you have, and times to let go with grace. Sometimes the right thing to do is to throw out the old schools of thought in the name of progress and reform. Sometimes the right thing to do is to sit and listen to the wisdom of those who have come before us. How will you know what the right choice is in these crucial moments?

You won’t.”

That is the entire endeavor. Sometimes you follow Lynch. Sometimes you invert him. The wisdom lies in knowing that no single piece of wisdom is final. You assess the environment, you judge the context, and you make a choice.

You won’t know if it’s the right one until later. But you have to choose anyway.

I’ve been writing here for almost two years. You may have noticed there was no celebratory milestone email, no confetti, no “look how far we’ve come” post. I wouldn’t know how to write one without sounding like an idiot. So just: thank you all for reading.

But somewhere along the way I noticed a pattern: a decent number of readers fundamentally misunderstand what I’m doing here. If many people keep asking the same questions, the common denominator is probably me. I didn’t explain myself well, or even at all. I just naively assumed people would understand. So here’s an attempt at a proper manual.

The feedback is almost always well-intended, but it often misses the point. (Occasionally I’m also told I’m a Chinese spy spreading propaganda, which is less well-intended but at least more creative.) You should build a brand. Show your real name. Why no photo. Use the Substack chat feature. Launch a podcast. Do YouTube. Display your portfolio. Give target prices. Send real-time trade alerts.

I’m not going to do any of that.

If you need a brand, a headshot, or a chat feature to decide whether something is worth reading, I have to warn you early: this will be disappointing. I’m not hiding because of some dark secret — have I mentioned I am bald? I just enjoy a peaceful life where my wife is not regularly informed by strangers on the internet that her husband is an idiot. She knows that already; she doesn’t need external confirmation or reinforcement.

More importantly, if an argument only works for you after you see my face, or after I give you a precise price target for a specific date, we’re no longer talking about research. We’re talking about marketing theatre.

I am bad at theatre.

Why I write at all

This might be disappointing, but I write for selfish reasons.

Long before Substack, I was already reading filings for fun, listening to too many earnings calls, digging through Chinese regulations, doing scuttlebutt.

MIT likes to say studying there is like drinking from a firehose.

That’s how I do research. Absorb everything, then distill. Writing is that distillation. It forces precision. In my head, everything sounds brilliant. On the page, half of it turns into nonsense. The gap between those two is where the actual learning happens.

For years this lived in a private journal. Then Covid happened. I was locked down in China, and started putting parts online to see if anyone cared. A few people did.

Great people did! Some knew far more than I do — former employees, local operators, fellow investors who’ve watched a sector for ten years longer than I have. They told me where I was wrong and where I was missing data. They engaged with me.

From the outside, you barely see any of that. The comment section looks empty. The interesting part happens in private. For me it’s pure upside: writing makes my thinking less bad and attracts people who can challenge it, while still letting me keep a low profile in real life.

Who this is not for

If you want a model portfolio you can copy, real-time trade alerts, target prices, neat DCFs, or personalised investment advice — this is the wrong place. Those are perfectly reasonable things to want. Plenty of smart people provide exactly that. I’m not one of them.

Have you noticed that when people say “this is not investment advice,” what follows is usually exactly investment advice? When I say it, I mean it. No targets, no alerts. Just ideas, insights and observations you can use or ignore.

I’m genuinely humbled that a very small number of people have asked me to manage money for them. But I’m not going to run a fund or manage anyone’s capital.

There are many good reasons, but the only one that counts is this: doing that would destroy what I care about. My freedom would shrink. I report to nobody. I don’t own a calendar — if I have an appointment, it’s important enough that I can remember. Otherwise I don’t have that appointment. I wake up in the morning and decide what to work on. I like it that way.

The moment I manage outside money, I start writing defensively, softening language, defending positions that might look stupid at a dinner table. And I really don’t want phone calls from nervous investors telling me a stock I own is down 25% — I can read a screen. I’d rather spend that time figuring out what actually changed than playing therapist to someone’s brokerage app.

I don’t hold hands.

I prefer the current setup: I absorb as much as I can and stay focused on research. I manage my own capital. You manage yours. We meet in the middle at the level of ideas.

There is exactly one standing rule: if I write positively about a company, I own it. This doesn’t mean I’m right about it. It just means I’m eating my own cooking, for better or worse. But I’m not telling you my position sizing, when I add, when I trim, or when I sell. I don’t write about all the companies I own.

How I think about stocks

After some of my recent articles, I got strong feedback that made me realise I never properly explained this part. My fault — but this Substack was much less planned than you may think. If you come from a world where every opinion must be either “all in” or “this is trash,” my writing will frustrate you.

I have no emotional attachment to any company I own. I don’t feel attacked if you think something I like is uninvestable or a stupid idea. I don’t need external confirmation — I only mention that to sound humble, since I just told you not to reinforce my wife’s views.

Disagreement is the only information that can actually change my mind, which makes it the only information I care about. I’m not team bull, not team bear, and I don’t join cults around tickers or founders. The founders certainly don’t know I exist, so it would be a very one-sided cult. That said, Pony Ma, if you’re reading this — may I meet you? (I’m only human. Sorry for this moment of weakness.)

In practice, this means: when I look at a company, I try to identify the three or four things that really move the equity — key business drivers, not every KPI under the sun. If I can’t get the story down to a handful (sometimes just one) of variables, I probably don’t understand it well enough to own it. Then I assign probabilities to different scenarios for each driver and keep updating those as real data comes in. Most headlines are noise. A few datapoints actually move the odds. Those are the ones I care about.

This creates a problem when you only see the written output. If new information makes bad outcomes more likely, I’ll write a critical piece. That doesn’t necessarily mean I sold, or that I “hate” the company now. It means the weight I give to certain negative scenarios has increased, which changes expected value. From the outside, you see the negative tone and it looks like a U-turn. From the inside, it’s just a probability update.

There’s a second distortion. Because I already know the upside story for anything I own, I rarely spend space repeating it. I use writing time to stress-test what can go wrong. That makes the public tone systematically more negative than my actual portfolio. If you translate everything I write into a simple buy/sell label, you will keep misunderstanding what I’m doing.

I focus on the downside because the upside doesn’t need my protection. If something wonderful happens that I didn’t foresee, I’ll take the money. If something terrible happens that I didn’t foresee, that’s a failure of imagination I have to own.

Who this is for

The way I hope you use my Substack is how I interact with the people I learn from: as a sparring partner. I read and discuss with others to test my hypotheses, steal useful datapoints, and see where my story breaks. Treat me the same way. Take what’s useful, compare it to your own work, see whether anything changes in your head. Ignore what isn’t.

If you can hold in your mind that a company can be attractive overall and still have serious risks whose odds shift over time — you’re probably in the right audience. If you can read sharp criticism of a business you own without treating it as an attack on your identity, and you want to check your thesis against someone who spends an unhealthy amount of time reading about these companies, you’re probably in the right place.

If you mainly want a simple buy stamp, a price target, or the next guaranteed ten-bagger — wrong address.

My tone and attitude will stay the same: low tolerance for corporate euphemism, zero interest in fan clubs, and a tendency to say things not optimised for everyone’s feelings.

If my tone bothers you more than weak governance, or more than overlooking a glaring hole in an investment thesis, this probably isn’t the right place for you.

If You’re Still Here

Rereading this, I worry it sounds like I have a secret identity. I don’t. I’m not Batman — no butler, no cape, no Batcave. Quite a few people know who I am, and several have been in my apartment when they visited Shanghai to enjoy the view, drink coffee, and discuss stocks for hours. Let me know if you are in town.

I started writing for selfish reasons, but somewhere along the way it became something bigger. The messages, the people I’ve met through writing — that’s the best part. I’ve met so many amazing people. Some are still in college. Others are already retired. Some are like me, unknown to everybody. A few are well known in the investing world — people I looked up to when I started investing. And now, through this writing, I get to exchange ideas with all of them. I still can’t quite believe that.

I didn’t start this with a strategy. I just started. And that turned out to be enough. If you’ve been waiting for the right moment to share your ideas — stop waiting. The people who need to hear them will hear them. And some of those people will change how you think. That’s the whole magic of it.

I recently laid out my thinking on DiDi and what I learned from my discussion with investor relations.

Then Q3 results came out, and I thought a brief update was in order.

Reading through the quarterly report, I was immediately reminded of Wolfgang Pauli, the Nobel Prize-winning physicist famous for dismissing bad theories with the ultimate academic insult: “That’s not right. That’s not even wrong.”

My first instinct looking at DiDi’s numbers was similar—”That’s not good. That’s not even bad.”

Having sat with it longer, I reconsidered. It’s not that the results exist in some quantum state below bad. They’re just bad. The China business is actually okay—mildly good, even. But some of the concerns I raised in my last piece are actually worse than I initially thought.

But I’m getting ahead of myself.

This article is for paid subscribers. You can read the China section of my latest DiDi piece for free — and if you enjoy it, consider becoming a paying member to unlock the full analysis.

Read more

Most companies still treat earnings calls like amateur theatre. The standard script is predictable: minimise the problems, pretend the risks don’t exist, recycle a few blue-sky fantasies, and hope investors nod along. Some do it with impressive confidence.

Didi once explained to me that they would take 30% of Brazil’s food delivery market, apparently without any chance of burning billions or failing outright. Baidu proudly announced in their Q2 2025 call that ALL robo-taxis were now safety-driver-free, a curious achievement considering that one week before that in Shenzhen EVERY single car I saw had a human behind the wheel.

JD keeps overselling itself to a level that would be funny if it weren’t so misleading. They repeatedly claim broad consumer mindshare across all categories, when in reality that “mindshare” only exists in electronics; everywhere else, JD is an unimportant player in China’s e-commerce market. And the latest gem was their insistence that their food-delivery business will ultimately be profitable as a standalone entity.

Our goal is to create a sustainable business that drives healthy order growth and at the same time, gradually and long scale effect and enhance operations with better UE. Ultimately, JD food delivery should be a self-sustaining business.

JD, Q3 2025 call

With a market share of roughly 10% against Meituan and Ele.me both roughly 45%, that is a structural disadvantage of roughly 4.5 to 1. Businesses with that kind of scale gap almost never reach sustained profitability — just ask Ele.me, which never managed to escape its disadvantage (70/30 marketshare for Meituan vs. Ele.me) before Alibaba went all-in on food delivery.

And then there is Pinduoduo the strange outlier: the one player that insists on highlighting what could go wrong. No victory laps, no fantasy projections. Their international expansion could easily be framed as a “massive opportunity across multiple continents.” Instead, they talk about “heightened uncertainty,” “regulatory unpredictability,” and “volatility in outcomes.” If JD owned Temu, we all know the script would sound very different — including dozens of mentions of “consumer mindshare.”

Subscribe now

I’ve been thinking a lot about why Pinduoduo behaves this way, and a few weeks ago the pieces finally fell into place. A conversation with one of ByteDance’s early investors, my wife reading a book about Duan Yongping — Colin Huang’s mentor — and several smart investors who pointed this out to me long ago made me realise I had completely underestimated a long-term tail-end risk for Pinduoduo.

But before getting to that, a quick look at the earnings. Nothing in the numbers should surprise anyone who understands what is actually happening. What is far more interesting, as always with Pinduoduo, is what they don’t say — what is happening behind the curtain.

Read more

Three things in life are certain: death, the second law of thermodynamics, and Tencent growing ad revenue by 20%.

Though honestly, with recent breakthroughs in longevity research, quantum computing, and AI-driven drug discovery, death and entropy might become optional. But Tencent’s 20% ad growth? That’s starting to feel like a new law of physics.

What continues to amaze me is the sheer farsightedness of Pony Ma and his team. People have already forgotten, but in December 2022, Ma held a pivotal internal meeting where he made a stark strategic pivot. He declared short video “the hope of the whole company” and turned it into Tencent’s top priority going forward.

This was Tencent’s Bill Gates “internet tidal wave” moment. The kind of strategic recalibration that separates great companies from roadkill. And in hindsight, Pony Ma and his team have been completely right. But more on that later.

Share

What makes this pivot particularly impressive is that Tencent had already tried and failed in short video. Twice. Remember Weishi (微视)? Or Vue Vlog? At one point, they did what they often do when they can’t crack something themselves: they invested in another short video player — Kuaishou. But Pony Ma and his team realized how critical short video would become. So they sat down, studied what went wrong, absorbed the lessons, and came back with Video Accounts. This time properly integrated into the WeChat ecosystem. That kind of strategic persistence, learning from failure rather than running from it, is what distinguishes them.

I’ve been thinking a lot recently about business models—AI infrastructure, e-commerce in the age of AI, social commerce, ad tech. And suddenly, the connective tissue became visible. Things are falling into place across Tencent’s ecosystem in a way I hadn’t fully grasped before. I underestimated Tencent again. I now see much more clearly where they’re heading. And when the pieces align like this, it’s scary for their competitors.

Read more

If you’re an investor in Alibaba or have traveled to China, you’ve probably come across Amap (高德地图, also known as Gaode). It’s a mapping and navigation app, basically Google Maps except it actually works in China.

Alibaba is now using Amap as a central weapon in its push into Meituan’s stronghold of instant retail and food delivery. It’s worth understanding how Amap fights and wins.

Founded in 2002, Amap started life as a boring enterprise mapping software company. The kind of business that makes money by selling map data to car manufacturers and government agencies.

By 2014, when Alibaba acquired Amap, the company was in deep trouble. Baidu Maps was destroying them.

Subscribe now

When Baidu Was Actually Good at Something

Here’s the situation Amap faced in 2014: Baidu Maps had a staggering 65.2% market share. Amap? A pathetic 20.8%, less than a third of Baidu’s dominance.

Baidu had every advantage:

Search was the gateway to maps

In the PC era, people didn’t just open a map app, they searched for directions on Baidu, which helpfully funneled them straight to Baidu Maps. It was the perfect traffic funnel.

Baidu had money

Lots of it. It was still printing cash from search advertising, before deciding to diversify into, well, everything else it’s now mediocre at or has quietly abandoned. Back in those days, BAT actually meant Baidu, Alibaba, and Tencent — not ByteDance, Alibaba, and Tencent. At the peak of that era, Baidu’s market cap was roughly on par with Alibaba’s. So Alibaba had to compete on equal terms and couldn’t simply outspend Baidu.

Network effects were already kicking in

More users meant better traffic data. Better traffic data meant more accurate routes. More accurate routes meant more users. The flywheel was spinning, and Amap was getting left behind.

In any rational analysis, Amap was done. You don’t come back from a 3:1 deficit against an entrenched leader with deep pockets and a built-in distribution advantage. Except... they did.

Enter Yu Yongfu: The Exception to Every Rule

When Alibaba completed the Amap acquisition, they needed someone to turn this around. They chose Yu Yongfu, who had just joined Alibaba through the UC Browser acquisition.

Here’s something you need to understand about Alibaba’s partnership structure: becoming an Alibaba Partner typically requires over five years at the company. It’s their equivalent of making partner at a law firm. It’s a big deal, lots of vetting, proof of loyalty and competence.

Yu Yongfu made partner in one year.

The only person to ever do this. Either Alibaba’s leadership completely lost their minds, or Yu Yongfu is genuinely exceptional. It was the latter.

How to Beat a Rich Competitor: Stop Playing Their Game

Despite Baidu’s huge percentage lead, the market for navigation was nowhere near saturated. The fight wasn’t just about stealing users; it was about capturing a vast wave of new users who had never used a mobile navigation app before. While Baidu assumed the race was already over, Yu saw that the track had only just been built.

He also realized that the shift from PC to mobile fundamentally altered how people used maps. On PC, you searched for directions. Baidu owned search, so they owned the entry point.

On mobile, you navigate in real-time. You’re in your car, or on your scooter, or walking to lunch, and you need turn-by-turn directions right now. Search isn’t the entry point anymore, the map app itself is.

This destroyed Baidu’s structural advantage. Suddenly, it didn’t matter that Baidu had the best search engine. Mobile users opened map apps directly. The playing field was level.

And Baidu, being Baidu, completely missed this shift. Baidu Maps was already in monetization mode, squeezing its existing user base long before the product was strong enough to lock in its advantage (the same pattern they repeated years later when they rushed to commercialize their Ernie bot)

At the same time, they began stuffing the app with O2O services that no one actually wanted inside a navigation interface, diluting focus and adding friction to the one task the product existed to perform.

Yu looked at them and said: fine, you do that. We’ll build the best navigation product in China.

Share

First, he declared Amap wouldn’t monetize for three years.

He started by declaring that Amap would not monetize for three years. Let that sink in. You’re losing badly, your rival controls the market, everyone’s panicking about losses, and you tell the board you’ll make zero revenue for three more years. A bold career choice.

Second, he cut all the O2O nonsense.

Then he cut all the fluff. The mid-2010s were the golden age of Chinese super apps, when every company decided they had to do everything. Movie tickets, hotel bookings, food delivery, astrology if necessary. Baidu bought into the fad. Amap had tried it too. Yu killed it. He stripped Amap down to the essentials: navigation and maps. Nothing else. If you can’t guide a driver from A to B faster and safer than anyone else, why exist?

Third, he targeted the people who actually matter: professional drivers.

Most navigation apps chase the mass market—more users, better data, faster flywheel. Yu inverted the formula.

Professional drivers, taxi drivers, truck drivers, delivery couriers, ride-hailing drivers, use navigation constantly, often spending eight to twelve hours a day on the road. If Amap could win this group, it would gain the most valuable real-world traffic data, since these drivers cover far more ground than casual users.

It would also gain organic, high-frequency exposure, because every passenger sitting in a taxi or bus sees which navigation app is being used. And most importantly, professional drivers are the most demanding users of all; if you can build a product that satisfies people who navigate for a living, winning over everyday users becomes easy.

So Amap invested heavily in the features that professional drivers actually need. Better lane guidance. More accurate traffic data. Services to help drivers find bathrooms or cheap, good restaurants along their routes.

Cultural growth accelerant: celebrity voice packs

When Amap introduced voice navigation featuring Lin Chi-ling (台湾第一名模, basically the most famous model/actress in the Chinese-speaking world at the time), suddenly everyone wanted to try Amap. Downloads spiked. People actually switched apps just to have a celebrity voice give them directions.

Was this a sustainable competitive advantage? Of course not. But it got people in the door. And once they were in, they discovered that Amap’s navigation was actually... better.

The voice packs brought the users. The drivers brought the data. Together, they tipped the market.

The Compounding Power of Better Data

All of these decisions and insights were an overwhelming success. By 2016, Amap had overtaken Baidu Maps, and their market positions had essentially reversed. Baidu had fallen to a distant second.

Amap’s turnaround came from reversing the data flywheel. Once it won professional drivers, its real-time traffic data improved quickly, producing better routes, which attracted more users, which further improved the data. The loop that once favored Baidu now spun for Amap. Crucially, Amap kept investing heavily in data collection—lane-level mapping, live traffic feeds, POI accuracy, and integration with road systems—while Baidu cut back and tried to squeeze profits. By the time Baidu realized it needed to fix navigation quality, Amap’s data advantage was so large that catching up would have required years of sustained investment. The network effects and user habits were already locked in, and Baidu never regained ground.

From Navigation App to Super App (But Smartly This Time)

Remember how Yu Yongfu cut all the O2O features? That was true, for a while.

Once Amap had firmly established itself as the dominant navigation app, once the core was unquestionably better, they began adding services back. But this time, the expansion was deliberate, not frantic.

Ride-hailing integration came first: if you’re navigating somewhere, you may need a ride. This wasn’t a distraction—it was a natural extension of the navigation use case.

Then came travel bookings via Fliggy. If you’re planning a trip, you can book hotels and flights directly from Amap. But importantly, Amap didn’t try to build a travel service itself. It provided the traffic and discovery layer, while Fliggy handled the actual booking infrastructure and supply.

This division of roles was smart. Amap gained utility; Fliggy gained high-intent users. Both sides benefited.

The key point: Amap only expanded after it won the navigation war. It didn’t dilute focus or confuse the product’s identity. It built outward from a position of strength, not desperation.

Subscribe now

Conclusion

The history of Amap is a masterclass in strategic focus. But the reason to study it now is that Alibaba is attempting to repeat the playbook, this time against Meituan.

Meituan’s position is formidable, built not just on a delivery app but on a decade of user-generated restaurant reviews on Dianping and a deeply integrated merchant supply chain infrastructure through Kuailv and a hyper-local logistics network. Alibaba isn’t just pushing Ele.me and Taobao to compete head-on. They are using Amap as the new point of entry. Recent feature rollouts are transforming Amap into a local discovery platform that overlaps directly with Dianping’s core territory. There are some interesting moves happening right now, which I’ll break down in more detail when I cover Alibaba’s next earnings.

The cycle repeats. Meituan has the lead, but the instant retail market is still growing. Alibaba appears to be executing the same playbook: grow the market, delay monetization, build the best product, and let the compounding effects of network and data take hold. Amap was once the underdog that overturned a dominant incumbent. Now, Alibaba is betting it can be the weapon to do it again. If it works is another question.

Let’s start with a confession: I can’t imagine my life without DiDi. I love their service. They’re the very reason I sold my car, and I’m not thinking for a minute about buying another one. I like the business too, I like the whole setup with trading over the counter and having the option of being relisted in Hong Kong. I’m the second-highest tier user, and everyone on my company’s staff uses their service.

I’m giving you this background so you understand that what follows comes from a place of affection, not animosity.

But a couple of weeks ago, I had a conversation with DiDi’s investor relations team that left me feeling... let’s say uneasy. The kind of uneasy you get when someone you trust tells you not to worry about something that absolutely deserves worrying about.

To be clear, Didi’s investor relations team was fully professional and genuinely helpful — which, in fairness, isn’t something one can take for granted when dealing with Chinese companies. My concern isn’t with the IR team itself, but with what was actually said.

I’ve been mulling this over ever since, but because DiDi updates are among the most requested from readers, and because some news I read a few days ago finally convinced me to put this in writing, here we are.

Subscribe now

China business: Going Strong

Didi’s core business the China business looks good. I have written about it before. Take rates are improving. EBITDA margins are heading in the right direction. User stickiness remains strong. Market share is defensible. Competitors like Caocao (the number two player 5.4% market share) have done precisely nothing of consequence. In fact, according to Caocao’s (HK:2643) own IPO prospectus, DiDi commands roughly 70% market share. That’s the kind of dominance that comes with serious scale advantages and network effects. Even the aggregators like Gaude (Alibaba’s Map service), while theoretically a threat, seem more like background noise than an existential crisis at the moment.

There’s still significant room for growth in lower-tier cities. IR told me that even in mature markets such as Beijing and Shanghai, they’re still seeing 6–7% year-over-year growth in trip volume, while second-tier cities are growing around 10%, and lower-tier cities even faster, as they continue to gain market share from traditional taxis. Compared with international peers like Uber and Lyft, DiDi’s share of total trips remains much smaller, leaving ample space for expansion.

From what I’ve observed firsthand in recent years—speaking with people across different demographics—the social status once associated with owning a large car has declined sharply. As a result, more users are likely to shift to DiDi, since on a pure cost basis, using DiDi is far cheaper than owning a car once you factor in parking fees, maintenance, repairs, and the many hidden costs of ownership. IR told me that in major cities such as Beijing and Shanghai, they estimate that using Didi costs about 30–40% of what it would cost to own a car — a figure I find entirely believable.

Didi is not only cheaper than owning a car, but also far more convenient — you don’t have to drive through traffic yourself or waste time finding parking; the driver simply picks you up wherever you are and drops you off right at your destination.

As of now, I’m fairly optimistic about the China business. Management expects around 10% volume growth next year, along with slightly better margins, which seems reasonable to me. If nothing major changes, I’m confident DiDi will maintain stable growth—perhaps not at the high rates of the past—but continue to dominate the field while steadily improving its margins.

The Ride-Hailing War Is Over (Probably)

It appears the great ride-hailing war in China has reached an end. The battlefield has been surveyed, the casualties counted, and everyone seems content to hold their territory. So the market leader should now enjoy its scale advantage.

Except… I’ve thought this before.

I thought the same thing about Meituan’s food delivery dominance. Then JD and Alibaba decided to throw logic out the window and burn billions of dollars to gain market share. I genuinely did not expect rational economic actors to behave that irrationally. But it’s China—and they did. And they might again.

The saving grace for Didi is that ride-hailing doesn’t seem to be the strategic focus of any of the big boys. Food delivery? Sure, that’s core if you’re already in retail. It’s adjacent, it’s logical, it’s maybe worth a subsidy war. But ride-hailing sits in a different category altogether. Even Meituan, which once tried to challenge DiDi, has pulled back.

Still, this remains a very China-specific risk—one that DiDi’s management simply can’t prepare for if another player decides to throw rationality out the window. I mention this because it’s something one must always keep in mind and watch closely in China. For now, though, there’s no sign of irrational competition.

That said, I’m confident in the China business. It appears stable—for now. Which, in Chinese tech terms, means “probably fine until it very suddenly isn’t.”

I’ve covered the bright side. What follows is the part that makes me nervous.

Read more

One of my core investing principles is obsessing over the value of information. I’d estimate 95%—probably more—of what I read is pure noise. Most of it won’t matter in one moth let alone six months. Most of it exists solely to make me do something stupid.

So whenever I read something, I ask: Does this actually matter? Or is this the financial media equivalent of empty calories—filling, forgettable, and potentially harmful?

More information isn’t better. The right information is crucial and the timing matters. I actively label things as wrong or useless so they don’t accumulate in my brain like some kind of intellectual junk drawer. It takes effort. But I’d rather remember three things that matter than three hundred things that don’t.

My favorite quote from 100 to 1 in the Stock Market by Thomas Phelps is:

When you read a bearish story on a company whose stock has declined to a third of what it was two or three years ago, ask yourself not only whether the story rings true but also why it was published at this late date. It may be factual but still highly misleading to investors because of its timing.

Factual doesn’t mean valuable. Timing changes everything.

Subscribe now

When Everyone Agrees, Ask Why Now

This brings me to the sudden flood of AI capex warnings—articles arguing that tech companies may be overspending on data centers and infrastructure to power their AI ambitions. The concerns aren’t new; I saw early articles on this more than a year ago. But lately? They’re everywhere.

Some are genuinely well-researched pieces, I very much like the two recent pieces by Harris Kupperman from Praetorian Capital. But the sheer volume, and the timing? Suspicious.

Even Jeff Bezos recently said at a talk in Italy that artificial intelligence is currently in an “industrial bubble” where both good and bad ideas are being funded. However, he added that the technology is real and will ultimately bring significant benefits to society.

The question for an investor isn’t whether the warnings are factual, but whether they are valuable information or just well-timed noise. To answer that, you need a framework.

AI Will Change Everything. That Doesn’t Make It a Good Investment.

I personally think AI will have a transformative impact on society—probably as significant as, if not greater than, the internet. There’s no question about it. The technology is real.1

But that doesn’t mean you should be the one funding the buildout. The railroads changed America. The fiber-optic networks power everything we do online today. Both were essential infrastructure. Most of the companies that built them went bankrupt anyway.

The question isn’t whether AI matters. It’s whether the companies burning billions on capex (and opex) today will be the ones capturing the value tomorrow—or whether they’re just laying expensive groundwork for someone else’s profit.

From Moat to Treadmill: Why AI Is Different

Everyone uses the same analogies—railroads, fiber-optic cables, shale drilling. I just used them myself. But the details matter, and they are exactly why I don’t like the comparison.

Railroads and fiber-optic cables, in theory at least, build a moat. Once you lay down tracks or bury fiber, nobody else is doing it. It would be economic suicide. This infrastructure lasts. In Germany, we still use train tracks and signaling from the pre-WW1 era. Fiber cables are the same. Once they’re in the ground, nobody’s ripping them out to build “better” fiber.

AI is different. What you build today is obsolete in 3 to 5 years. New chips, new architectures, faster models—they make last generation’s infrastructure worthless.

This creates a trap. If you stop investing and your competitors don’t, everything you spent becomes useless. You’re not building a durable moat. You’re running on a treadmill that speeds up every year.

Worse: rapid depreciation means any latecomer with fresh capital can deploy newer architectures and close the gap without first wasting billions on infrastructure that’s already obsolete.

This is structurally identical to the prisoner’s dilemma: mutual restraint would benefit everyone, but individual incentives guarantee an arms race. If you keep spending, you burn capital on depreciating assets. If you stop and others don’t, you fall behind. The only winning move—coordinated restraint—is unavailable because no one can trust competitors to hold back. This makes the AI buildout even more brutal than historical infrastructure booms.

Subscribe now

The Hyperscaler’s Defense: Ecosystems and Cash Flow

On the other hand, some of the companies building AI infrastructure today aren’t like railroad companies. Microsoft, Google, Amazon, Tencent—they aren’t pure infrastructure plays. They have massive cash-flow engines elsewhere. They can subsidize AI to reinforce broader ecosystems, even if AI itself runs at break-even or a loss.

I’ve written about this before with bike rental businesses in China. Meituan, Alipay, Didi—they all run bike rentals that barely break even, maybe lose money. But the point is bringing people into the ecosystem.

The difference? Bike rentals are tiny compared to the overall scale of these businesses. The losses don’t matter. AI capex? That’s serious business. We’re talking hundreds of billions. That’s not a rounding error.

For AI-first startups like OpenAI, with minimal revenue and massive burn, the bearish case writes itself. They’re spending far ahead of any proven business model.

But the picture is more nuanced than the blanket warnings suggest. The noise of capex concerns fails to make this critical distinction. Details matter, and the details vary wildly. A blanket warning about AI capex is about as useful as a blanket warning about tech stocks—which is to say, useless.

I went through all of my AI-related holdings and analyzed other players in the field, comparing their capex, opex, revenue trajectories, and whatever AI-related financial data I could extract. The picture is anything but uniform. For some I’m deeply pessimistic. The math simply doesn’t work. Pessimistic doesn’t mean doomed—I just can’t see the path yet. My strategy is to track them closely and wait for the fog to clear, then identify who survives and wins big, because there will be winners in that group. For others, it looks less like reckless spending and more like a well-calculated gamble with asymmetric upside.

To see if the noise has merit, let’s take Tencent as a case study. I went through their filings, management statements, and reliable online information.

A Case Study in Cutting Through the Noise: Tencent

Read more

Forget the Communist Party — it’s the capitalists you should fear!

One of the biggest misunderstandings among foreign investors is believing that China’s political system is the main reason its stock market keeps underperforming the U.S. It isn’t. Among many other factors, the real challenge, the one that quietly destroys profit margins and wears down investor patience, is the brutal, endless competition

Competition in China doesn’t stop when it stops making sense. It keeps going long after logic and balance sheets have given up. And it’s not only about price wars. The moment a business model starts working, it gets copied, cloned, and overextended until the entire market is flooded. Success invites replication, not consolidation. Everyone rushes in, everyone overspends, and the cycle repeats.

You can see it today in instant retail, before that in ride-hailing, group buying, livestreaming — and tomorrow it’ll be something else. The defining feature of China’s economy isn’t its political structure. It’s the reflex to compete past reason.

And this isn’t just about billion-dollar tech platforms. Walk down any street in any Chinese city and you’ll see the same dynamic on a smaller scale. Your local beef noodle shop has three competitors within 100 meters, all racing to the bottom on price while stacking loyalty programs and mini-apps.

This constant knife fight means that, unlike most markets where one or two players eventually dominate, China rarely settles. It doesn’t consolidate. It multiplies.

A Market with Too Many Winners

Take e-commerce.

Anywhere else, you’d end up with one or two major survivors. In China, we have at least seven.

China’s e-commerce battlefield now includes the three traditional players, Alibaba, JD, and Pinduoduo; the live-streaming platforms born from the social short-video universe — ByteDance, Tencent, and Kuaishou; and the instant-retail entrant, Meituan. And then I haven’t even mentioned Xiaohongshu, or RedNote as it’s known in English, along with many others. Xiaohongshu is now also aggressively expanding into e-commerce.

Subscribe now

AI Won’t Flatten the Playing Field — It’ll Tilt It

Investors, used to a single dominant platform at home, often fail to grasp the diversity of China’s e-commerce landscape. They lump everything together — Pinduoduo as a “cheap Taobao,” JD as “Taobao with warehouses.” That kind of shorthand might feel convenient, but it’s a fundamental mistake.

These companies aren’t small variations of the same model. They’re structurally different businesses with different incentives, strengths, and strategic logic. The nuances matter.

And with AI entering the picture, those differences matter even more.

I’m a true believer that AI will transform everything — probably faster and more deeply than most expect. But that’s exactly why investors need to get the foundations right. AI won’t reshape all these companies the same way. Their models will absorb, resist, or amplify it differently.

Lumping them together is dangerous lazy thinking.

Read more

Opportunities are born when fear overwhelms reason. Harvard’s Richard Zeckhauser pinpointed the dynamic perfectly: investors demand far higher compensation for risks where a “betraying human”—like a government—could be the cause of loss, compared to the whims of an indifferent market.

In my last article, I outlined Beijing’s playbook for sectors it deems misaligned with national interests. Yet the devil is in the details. The company/sector I will analyze today exemplifies a critical investor misperception: fearing a broad crackdown, many are dumping the entire sector, driven by that same aversion to governmental betrayal. Yet in reality, this firm operates within a subsector defined by radically different economics and enduring policy support. This fundamental misclassification has opened a compelling opportunity.

This article will dissect that opportunity. I’ll first examine the overlooked subsector to illustrate why its outlook is so divergent. The company itself trades at a mere 3.5 times earnings and offers an 8% yield. It has compounded revenue and profit for over a decade, while in reality, it remains strategically positioned to keep benefiting from government backing. The market’s myopic focus on short-term uncertainty has blinded it to these durable strengths.

Read more

Warren Buffett once said, “If I need a second opinion, I’ll look in the mirror.” It’s a reminder that investing isn’t a democracy. You don’t win by majority vote.

While I trust my own research, I highly value people who challenge my perspective, force me to rethink, and stress-test my assumptions. I generally only care about information that has the potential to change my current opinion —not reinforce what I already believe.

Recently, I had a call with a retired fund manager I got to know through my Substack. He’s become one of my key sparring partners for new ideas. I love discussing with him because he’s smart, not politically correct, and—most importantly—often holds a very different view from mine.

This time, I mentioned two Chinese companies I’m currently analyzing. In both cases, he didn’t even let me finish. Just said: “Beijing hates these industries. It’s dead money. Believe me.”

He generally has a much more negative view on China than I do, which I genuinely appreciate. But this time, it felt overdone. And for me, that was valuable signal. If someone smart won’t even hear the pitch, just hears the industry and shuts it down, that’s exactly where it gets interesting.

Because whether it’s euphoria, despair, or outright refusal to engage, that’s where mispricings hide. Always.

Subscribe now

Same Policy, Different Reaction

To give you some context, I was looking at two companies in sectors that have recently been hit hard by Chinese policy. So why were our reactions so different? He instantly dismissed them and refused to take a closer look—while for me, that actually triggered real interest.

It all comes down to perspective.

The Outside Perspective: The Demolition

To an external observer, China’s regulatory crackdowns look like arbitrary acts of destruction. The intervention is usually sudden and brutal. With little warning, policies land that wipe out valuations, declare entire business models no longer viable and trigger headlines announcing the death of a sector or industry. The impact is immediate and severe.

For people used to governments that move slowly, where policy comes through soft signals, drawn-out consultation, and careful transition periods, this feels extreme. Even hostile.

It paints the picture of a state dismantling its own economy. So the logic follows: if it looks like a wrecking ball and swings like a wrecking ball, the goal must be demolition.

The Inside Perspective: "The Government Is Not Stupid"

I’ve lived in China long enough to have internalized parts of how things work here. Still, I learn more every day—and there’s much more to learn. Just how different the local perspective is from the outside view became clear to me again during a recent call with the management and IR team of one of the companies I mentioned earlier.

They were very direct. They told me the entire industry is in a regulatory winter. The pain is real. Weaker players are being shaken out. But one executive said something remarkable: “It’s painful, but the government is not stupid.”

He didn’t say that out of party loyalty or some need to defend official policy. He meant it seriously. He understood why the government acted the way it did. Parts of the industry had developed in unhealthy ways—practices that didn’t serve any broader public interest. So the government stepped in.

Because in China, the purpose of these interventions is rarely to destroy a sector—unless we’re talking about the mafia or drugs, which did get wiped out in Beijing and other places. The point is realignment. When a sector—be it education, tech, or real estate—starts growing in ways that collide with bigger national goals, whether that’s social equity, financial stability, or demographic priorities, the state intervenes. They don’t spend two years consulting industry stakeholders and releasing white papers—they move. Fast. Publicly. Often with overwhelming force.

The purpose is twofold: to instantly terminate undesirable practices and to send an unambiguous signal to every market participant about the new inviolable rules of the game. It is a short, sharp shock designed to clear the field.

And when the executive said, “The government is not stupid,” this is exactly what he meant. Beijing knows that China very much needs the sector this company operates in. They know how important it will be going forward. Which is why, now that the initial impact has landed, they’re already starting to fine-tune the policies—quietly adjusting where they went too far.

Subscribe now

Pattern Recognition

What I’ve described here isn’t new. Variations of this pattern have played out many times before.

I’m not going to explain why the Chinese government acts the way it does. There may be others who can explain that better. That’s not the point.

What matters is the structure—the shape these interventions tend to take. Once you’ve seen a few of them, it starts to look less like chaos and more like a recurring pattern.

It usually follows a familiar outline—a kind of three-act play. And no, understanding the exact motivations behind every policy move isn’t the point. The point is recognizing what’s happening, where we are in the cycle, and what tends to follow.

It’s pattern recognition. And once you’ve seen it, it’s hard to unsee. Master that pattern—and you’ll find that China’s best opportunities often hide not in today’s darlings, but in yesterday’s disasters.

Read more

If you want the short version of China’s instant retail war so far, it goes like this:

Meituan ruled. JD pulled the moral card. Alibaba said, “hold my beer,” and lit the voucher fuse.

For years, Meituan owned food delivery. Ele.me was a distant, loss-making second. Then JD marched in with no infrastructure, no riders, no experience — just grand speeches about being “nicer” to drivers and restaurants, paying social security. It was moral theater with nothing to show for it. Wait did I say nothing to show for? They did manage to make all their earnings disappear.

Alibaba, meanwhile, did the grown-up thing. Instead of whining about fairness, they competed — on price, on service quality, on scale. Ele.me got wired in as the local backbone of Taobao Flash Buy. And now, with AMap, they’re aiming at Dianping — a vastly under-discussed app that is one of the biggest cornerstones of Meituan’s moat in offline discovery.

The result so far: three companies bleeding cash, one very entertained consumer base (free coffee, free drinks, thank you very much), and headlines reduced to a single lazy line: “everyone’s losing money.”

In this article, I’ll go deeper — into the mechanics, the ecosystem plays, and the cracks that don’t make the press releases.

Read more

Over the last months, I read dozens—if not hundreds—of earnings calls and quarterly reports. And I’ve come to one unavoidable conclusion: every company is now an AI company.

Funeral homes now claim to use AI to enhance the client’s journey, including the one inside the casket. Pig farmers deploy AI to monitor hog mood swings. And naturally, every tech CEO now opens with how their model outperforms every other model on metrics no human understands, for use cases no one asked for.

Every company?

Not quite.

Like in one of those Asterix and Obelix comics, there remains a small, stubborn village. Pinduoduo. In their most recent earnings call, they used the word “AI” exactly zero times. Not once.

Lately, I’ve been asked over and over again: why isn’t Pinduoduo participating in the AI race? In this article, I’ll look at their history, how they operate, and why this silence fits the very strategy that made them so successful.

Read more

Alibaba’s tone changed. So did their posture. And a few moves suggest the story might be entering a very different phase. But it’s not all good news. There were parts of the reporting I didn’t like—and I’m also concerned about some developments we’re likely to see in the third quarter.

Read more

It’s that time of the year again: Pinduoduo’s second quarter earnings. And once again, we’re treated to the by-now familiar ritual. They post the numbers — this time even beating analyst estimates, which, when it comes to Pinduoduo, usually have all the predictive accuracy of a horoscope. Nevertheless, the stock jumps 10% in pre-market, because apparently people still cling to these numbers — even when they mess with their heads.

And then comes the call. As per the sacred Q2 tradition established last year, management can’t help but remind everyone, multiple times, that these profits aren’t sustainable, that volatility should be expected, and that no one should get too comfortable.

And with their classic closing line, “I think it’s about time,” the curtain drops. Everyone goes home a little more confused. But they’ve achieved their goal: Colin Huang is still not the richest man in China. Mission accomplished.

Before PDDs Numbers

Before looking at the actual earnings, I want to start somewhere else. Because looking at Pinduoduo’s Q2 2025 in isolation — margin squeeze here, subsidy spike there — tells you very little.

Instead, I take a step back and examine where Pinduoduo has been, how it got here, and what I believe explains why management is doing what it’s doing — the shift in strategy, the spending, and the repeated insistence that profitability isn’t the goal.

I also look at the changing dynamics in Chinese e-commerce, some of the political pressure shaping the playing field, and how that influences the numbers.

As always, I focus on what I believe is overlooked or not talked about enough — not on what’s already been said a hundred times.

Read more

Last week I wrote that Baidu’s robo-taxis weren’t yet operational in Shanghai Pudong, despite their otherwise confident press releases. Here it is in case you missed it.

This week, I already rode in one.

Things are moving fast in China’s robo-taxi business. What was “not yet available” a few days ago is now casually U-turning across the street to pick you up.

For paying subscribers, I share my first-hand experience of Baidu and Pony.ai’s service. Which one felt smarter? What stood out? And how one subtle feature from Pony hints at what the future of autonomous fleets might actually look like.

I also discuss Baidu’s latest earnings and what it reveals about their position in the AI race.

If you’re not a paying subscriber, here are a few free articles to catch up:

Read more