The Secondary-Private Market: Underwater Deals, LLM Shares, and 321 Companies Begging to Be Bought

“I’ve said it countless times: the world before the Didi incident and the world today are two different worlds.” — Zhuang Minghao, on why China’s secondary-private market has to be told from June 2021

Guest: Zhuang Minghao｜host of the podcast The Art of Dragon Slaying (Tulong Zhi Shu), internet strategic investor, in VC since 2011

Five thousand dollars is enough to “invest” in SpaceX. A company that once required tens of millions of dollars just to touch has been sliced down, SPV layer by SPV layer, until the smallest unit costs five thousand. As we recorded this episode, another data point sat on the table: four days later SHEIN would list — a company once worth $98 billion at its peak, going public at a $26 billion offering.

The stretch of water between those two numbers is the secondary-private market — what the Chinese call the “1.5-tier market”: companies that haven’t listed, whose shares nonetheless change hands privately, at valuations that come neither from fundraising rounds nor from an exchange. Zhuang Minghao recently produced an episode called “The Underwater World of China’s Secondary-Private Market,” using AI to clean five years of 36Kr’s “Capital Sentiment Message Board” — 183 issues, 5,032 old-share trading leads — into a database that amounts to a micro-history of China’s private markets. He entered the industry in 2009 and has been doing real VC work since 2011; through every cycle of those five years, he was at the table’s edge.

Raymond has been in the water himself: pooling money with friends to buy old shares, setting up group chats, hiring lawyers, haggling endlessly — until he finally asked himself: what’s the point? So he came to this episode carrying skepticism: in pricing without financial statements, how much is fundamentals and how much is FOMO? This episode takes apart, one by one: the $5,000 SpaceX stake, the “SpaceX investors” in your WeChat feed, the real Mag7, Zhipu old-share SPVs, ByteDance the yardstick, SHEIN’s ride from 98 billion to 26, and 321 for-sale notices.

What follows is the full conversation, edited and condensed.

1. “A valuation that comes neither from fundraising nor from listing”

Raymond: For listeners who know nothing about this: what exactly is the secondary-private market?

Zhuang Minghao: The Chinese term literally means “tier 1.5,” sitting between tier 1 and tier 2. The primary market is simple: early-stage venture capital — investing in an unlisted company from angel rounds through late stage. The secondary market is the company going public: shares formally traded, with pricing, trading hours, and a full toolkit of financial instruments.

And the state in between? The company hasn’t listed, but at a certain stage, individual shareholders, co-founders, or the early funds that backed it for years want out, and the shares need liquidity. Some want to buy, some want to sell, so you need a place to trade, and both sides need consensus on price — no consensus, no deal. What emerges is a valuation produced neither by fundraising nor by listing, but by private transactions. There are buyers, sellers, assets, prices — and of course all manner of intermediaries who exist to make deals close. That is the secondary-private market.

It sounds small, irregular, without any long-cycle trend. But precisely because China’s and America’s primary and secondary markets have been through so much turbulence these past few years, it changed. Every fund in the primary market has a maturity date — when the date arrives, the books must be settled. Public markets can exist in perpetuity; funds cannot. Yet the exit environment gave nobody room to settle on schedule, forcing everyone to find a way out in the middle. That enormous external pressure took a market that had been underwater and wildly volatile and made it gradually more formal, more supervised, visibly growing, and increasingly impossible to ignore.

Raymond: Let me add an example. ByteDance was founded in 2012 — fourteen years ago now. Fourteen years far exceeds most funds’ lifespans — even its angel investors should have exited twice over by now. I think ByteDance is a market-proof company: it could list in any market, good or bad. But for its own reasons it won’t list today. And a great many investors are stuck exactly there.

Zhuang Minghao: Which is forcing something else into existence. I entered the industry in ‘09 and started doing real VC in 2011. In the early years you almost never saw an unlisted company voluntarily buy back employee options every year at a set price — that practice essentially didn’t exist, and was vanishingly rare among Chinese companies. The names everyone knows — ByteDance, Xiaohongshu, Kimi — are results of just the past few years.

Raymond: ByteDance genuinely earns a lot of money. The other two may have their own reasons.

2. “People genuinely believe the story Musk is building”

Raymond: Your episode has a killer punchline: SpaceX used to be a mega-asset requiring tens of millions or hundreds of millions from VCs, and now five thousand dollars buys you in. How did that happen?

Zhuang Minghao: SpaceX is extremely representative. It carries every buff of Musk’s entire track record, and that buff shows up as a valuation that keeps climbing on an ever-grander cosmic narrative. Individual investors don’t much care about the financials — revenue, profit, margins, not really. People genuinely believe the story Musk is building.

So a great many people genuinely want in. But under the old trading system, SpaceX isn’t listed and individuals had almost no path: you’d have to become an LP in a top-tier fund, and the capital threshold and connections required put that out of ordinary reach. Meanwhile this unlisted, extremely famous, extremely fast-growing company that everyone is happy to pay for just sits there, every day — and far too many people want a piece. So platforms appear, quote-unquote “trading methods” appear, serving quote-unquote “individuals” — including plenty of China’s freelance angel investors. The underlying logic is identical to retail investors chasing IPO allotments: Musk delivered on everything he ever promised, so of course everyone wants a seat at his biggest bet yet.

Push the topic one step further — setting hype aside, purely in terms of financial infrastructure, credit is owed to the web3 wave. It gave people ways to participate; even if the chain has legal defects and incomplete links, at least some hooks connect. Push that to the right moment in time, and it clicks into place.

Raymond: Let me paint the picture. SpaceX, founded in 2002, is this generation’s science-fiction object — but participating used to mean millions or tens of millions of dollars. Over the past decade, US financial markets produced wave after wave of “financial democratization,” and the ideology boils down to one sentence: why should the rich get richer buying the best companies while ordinary people can’t even participate? I did an earlier episode on Jarsy, titled “How to buy SpaceX at half price before it lists” — the moment SpaceX’s IPO news broke, prices on that platform doubled. It uses tokenization to slice SpaceX exposure layer after layer after layer, down to a minimum ticket of five thousand. I think it’s the perfect specimen — era, company, and individual, all three indispensable — the most representative artifact of America’s secondary-private market.

3. “Every account shows one dollar — but whose dollar is it?”

Raymond: When did Chinese money start buying SpaceX?

Zhuang Minghao: In my statistics this time, the top names — ByteDance, Xiaohongshu, SHEIN, SpaceX — surprised no one. But there was another: Musk’s brain-computer interface company Neuralink. It ranks top five, ahead of Discord, Cainiao, and DJI. Why?

The answer is the same. During SpaceX’s path to listing there were disputes over exactly this: Musk’s companies have all been wrapped in many, many layers of SPVs these past few years (editor’s note: Special Purpose Vehicle — a shell entity set up for a single investment; stack one SPV under another and a large allocation can be sliced into countless small tickets). Once an SPV layer is added, the end clients become small enough that the company starts allowing — no, not allowing, it simply can no longer control — who buys. Every account shows one dollar; whose dollar it actually is becomes very hard to know.

These past few years, a lot of dollars with Chinese characteristics have needed somewhere to go. That money might come from co-founders after their company listed, or from retired big shots of the old scene. Not rich enough for a family office, no way into the LP base of a name-brand dollar fund — but they too feel they must participate in this technological tide. Individuals can hardly buy equity directly, so SPV stacking it is. Is what they touched the second layer or the third? We don’t know. But the moment a trade completes, they can, in theory, say: I am now an investor in that company.

Raymond: Which is exactly what I keep seeing on WeChat and Xiaohongshu: someone I’ve never heard of, bio reading “SpaceX investor.” I can never tell whether they put in five thousand or fifty million. By your logic, if I spend a hundred dollars today — fifty on OpenAI, fifty on Anthropic — I can claim I’ve invested in both?

Zhuang Minghao: Yes, you can. There are plenty of token-based ways to buy now. Factually it isn’t wrong.

Raymond: Okay — this is truly not investment advice. It all sounds absurd; absolutely not investment advice. If you actually get scammed, there’s nothing to be done.

Zhuang Minghao: As I said, there really are too many legal gray zones and defects in the middle. Even a large public platform like Robinhood, pushing unlisted-company exposure these past few years, did it with tokens — and strictly legally speaking, problems remain.

Raymond: I want to dig into that. When Robinhood launched tokenized pre-IPO shares of OpenAI and SpaceX in the second half of last year, both companies immediately jumped up and said: this is not legitimate, we don’t recognize it. Robinhood’s logic was: the equity layer is the equity layer’s problem; I’m listing tokens, which is a token matter. But that second layer means what you actually bought is an IOU from Robinhood — a claim that someday, when the company lists, maps to a price. So what you bought is trust in Robinhood — that it won’t run, won’t renege. You did not buy direct ownership of those shares.

One more thing. Under the current US regulatory framework, Chinese nationals are effectively barred from investing in US AI companies. Have you heard any gossip on that front?

Zhuang Minghao: Let me start with my side. In the first half of 2026, during OpenAI’s $380 billion round, a lot of briefs and documents from such structures were floating around, and many of them carried explicit requirements on Chinese identity.

A while ago, a small web3-leaning outlet wrote about this when SpaceX listed: exactly how many China-linked funds had participated in SpaceX via SPVs — you can find them in the cap table (editor’s note: the equity register). After the piece ran, plenty of people contacted the author asking to have it taken down. The money wasn’t even large — some funds’ early tickets of twenty or thirty million dollars, nothing by the standards of web3 funds in their fat years. Even so, many did not want the exposure.

Raymond: Did it name specific funds?

Zhuang Minghao: Of course — possibly in pinyin. Not a name most people would recognize, but those who know, know.

Raymond: Understood. There really is a lot of money out there.

4. “This index only admits companies riding the crest”

Raymond: The US has the Mag7 — Nvidia, Google, Apple and the rest. Your deck argues those are the “fake Mag7,” and the real Mag7 are seven others: SpaceX, OpenAI, xAI, Anduril, Databricks, Anthropic, and Stripe — the seven largest unlisted companies in America. SpaceX is already a $1.8 trillion company; OpenAI and Anthropic are about to list, and they’ll ring the bell above a trillion for sure — the difference is merely how many trillions. Squeezing SpaceX and Tesla in and out of the seven sisters is fair game; the trillion-dollar club suddenly ballooned. What happened?

Zhuang Minghao: Just listen to the names: they embody the massive consensus of America’s private market over the past year or two — AI, defense, and throw in chips: those three themes. Essentially all the money, all the people, all the attention, all the pre-IPO possibility is concentrated there. And these companies are growing extremely fast — revenue, users, valuations all visibly climbing.

It works like the S&P 500 constantly kicking out bad companies and adding new ones, which is why it can keep rising — this index is the same: it only admits companies riding the crest of the wave, so it goes up like crazy. America has had plenty of unlisted-valuation league tables before — ByteDance and Ant were probably on them — but companies that have stopped appreciating, like Klarna, don’t get into this index. So the index rises in a rather “hollow” way.

Consider the tail of those seven, Stripe — a payments company that just bought OpenRouter for $7 billion. Its momentum is such that its conference gets mentioned alongside AWS’s, Google Cloud’s, even Nvidia’s — it is the payment infrastructure of AI, riding the wave completely. And Stripe has never clearly signaled an IPO, so hunger and trading activity for it in the secondary-private market run even higher. There’s a US quarterly ranking of old-share trading activity in unlisted companies: Anthropic is surely first now; last year it might have been OpenRouter; Stripe ranks very high even though its size isn’t the largest. What it represents is the sentiment of this exact moment.

Raymond: Presumably also because Stripe hasn’t raised in a long time — no formal round has priced it. Whereas OpenAI and Anthropic have massive real capital needs and keep taking money in.

Zhuang Minghao: There’s a very practical angle too. Remember the year before last, when Sam was briefly shown the door at OpenAI and then came back? One reading I have: every year, Sam helps a large number of OpenAI employees sell their old shares — it is one of his most important annual jobs. This year’s batch should already be done — sold at a very high price, seven-billion-plus dollars; last year’s buyer was SoftBank. Not everyone can do this: distributing gains at that scale and that valuation, in a compliant, legal way, is not something just anyone can pull off.

Raymond: Anthropic’s round in April or May this year was similar — effectively an employee tender (editor’s note: a company-organized window for employees to sell shares in bulk). Though I have no idea who at Anthropic was selling. That’s an interesting question in itself.

5. “Then it becomes: bet, or don’t bet”

Raymond: How much of this is FOMO, in your estimate? Here’s why I ask. In normal investing you have an enormous amount of information. If I buy Pinduoduo today, I know exactly how the company performed over the past eight quarters, and on the earnings call I can plainly feel the company’s utter indifference to its own share price — a value investor’s trap and graveyard, and I can see it, because the information is extremely transparent. But today’s companies are deeply opaque. Take Anthropic, about to list. Is its ARR net or gross? — Unknown. ARR means Annualized Recurring Revenue; how much of it actually recurs? — Unknown. After this year’s token-maxing (the model labs’ collective sprint for token consumption), some users churned; will that be disclosed? — Unknown.

So in this pricing, how much weight is FOMO? Fundamentals are certainly growing, no question, but the price grows fast too. Can you decompose it — how much FOMO, how much fundamentals?

Zhuang Minghao: I think the overwhelming majority is FOMO. It can’t be helped. On top of the ARR narrative — you don’t even need Anthropic. Zhipu listed early this year; by mid-last-year, a batch of Zhipu old-share SPVs had already appeared in China, and the financial data used at that point was still Zhipu’s 2024 numbers. Which is to say: nothing. There were no fundamentals to look at. But the price, if I remember right, was around a 16-to-18 billion RMB valuation in the documents. It is, unavoidably, a potential trade propelled by enormous emotion. As we joke: willing hitter, willing target — a bet’s a bet. That’s the game.

Of course, the intermediaries who need deals to close will produce a BP-like document, a data sheet. But that material is, first of all, certainly stale — possibly all public information you already had. If you treat this responsibly, as a genuinely serious investment transaction, those materials are certainly insufficient. So it becomes: bet, or don’t bet. Nothing more. Granted, that’s because these particular companies are still developing at breakneck speed.

Raymond: Like the saying — when the rocket is boarding, don’t argue about seats. Get on first.

Zhuang Minghao: Mostly, yes. But across the secondary-private market there’s a whole set of companies pointing the other way: no longer appreciating, fundamentals flat for a long stretch. With those, you genuinely must think about what you’re betting on and scrutinize carefully. If you pick up Klarna today, or Discord — are they bad? No. Revenue still grows, scale is real, position intact, maybe even new-story potential. And for these companies, the richness of information in the secondary-private market is already indistinguishable from a prospectus. Then what you’re betting on reverts to a traditional-craft investment decision.

Raymond: Investors can bet both sides anyway: for stability, buy Pinduoduo; for cosmic thrills, buy those. None of this is investment advice — you can already hear which companies I dislike.

6. “Most get written off; priced at cost”

Raymond: Let’s talk about the awkward companies. A while back I read the unicorn league table from Ren Zeping’s team — a global ranking plus a China-only ranking. Beyond the top ten — ByteDance, Ant, SHEIN, Xiaohongshu and such — many names past #20 haven’t been mentioned in years. And crucially the table never demotes them: a unicorn in 2015 is still a unicorn eleven years later.

Zhuang Minghao: Same story as IT Juzi’s list — “the vanished unicorns.” These companies are in fact a major reason the secondary-private market exists. They all had their moment of glory, appeared countless times in funds’ LP letters, and once ran wild through IRR calculations. But when funds mature and the books must be settled, those accounts have to be flattened. How to flatten them becomes extremely messy.

China has funds that specialize in this — taking over an entire fund’s positions (editor’s note: S funds, secondaries funds that buy fund stakes or portfolios of fund-held assets). When they evaluate targets, they price at cost: what you originally paid, plus a little — sometimes nothing — depending on asset quality.

Raymond: What is there to add? Let’s name no names — every fund in China is a great fund and everyone makes money — say I backed ten projects: one did 100x, home run, and the other nine are probably dead. Why would anyone take that package?

Zhuang Minghao: The one that “made it” may only have made it on paper. No exit, or exit deeply uncertain.

Raymond: But the other nine are definitely done for.

Zhuang Minghao: Of course. Counted as zero.

Raymond: Counted as zero and someone still takes them?

Zhuang Minghao: Because there’s no way to take positions individually — taking one alone would be a true old-share purchase. Many funds must settle everything at maturity, so it’s all bundled. It’s really an audit exercise: ten portfolio companies, each with different status, cost, price, fair value — do a holistic audit, then negotiate the package price. At that stage most positions hang at zero, all written off — as they should be. Only the few worth discussing get the debate: discount, or a small so-called premium. Friends of mine in secondaries say this is most of their job. The good, fully-settled assets need no work at all — like Zhu Xiaohu selling his Xiaohongshu stake a while back: no debate needed, the price is in a sense pre-marked. That’s a perfectly fair trade.

Raymond: Is this a bit like used-phone recycling? I sell you the whole phone — was 6,000 yuan new, you take it for 1,000 — you strip it, sell the memory, junk the rest on the spot.

Zhuang Minghao: Possibly, yes. If your condition is good enough, it might sell whole. Every variation exists.

Raymond: Let me ask about one more company: Cainiao. This Alibaba-system trade also keeps resurfacing.

Zhuang Minghao: It represents a certain specimen of the Chinese club deal (editor’s note: a deal in which multiple institutions invest as a syndicate). It was a club deal from the day it was born. Everyone who joined naturally hoped the club’s organizer would shoulder everything — but the organizer ran into other troubles these past few years and can no longer carry it. To his empire, Cainiao is too marginal. It’s as if everyone wants to dine at my house, but my chef has clocked out — fine, order takeout, make do. The holders must find a way. But it was born a club deal, so nobody takes it over — who would? Behind every one of these names are many stories, and every story looks different.

7. “Someone at SIG is retiring, so a block comes out”

Raymond: Pricing fascinates me. Take Xiaohongshu: it recurs in your deck and in 36Kr’s trade postings, its price shifting constantly with the business and the external capital environment. Any patterns you’ve observed?

Zhuang Minghao: The base layer is still company performance — that’s the core. Xiaohongshu first appeared on this board at roughly $12 billion; the last traceable price should be $38 billion, higher even than the ~32 rumored in the news around Zhu Xiaohu’s sale. No recent news, of course — the DST round was probably in the 200s.

So: first, it correlates with the business. Second, with trading sentiment — the market’s ambient mood, whether the IPO timeline looks near or far, the endless China-US mess. Third, with the balance of buy-side and sell-side volume at a given moment. Some companies took money at a particularly lofty height, and those funds mature at roughly the same time — a burst of sell orders lands within a short window, and the price simply cannot be good.

And these things are hard to negotiate. China has no scaled online platform like America’s where you can look up prices; much of it is haggling: a roughly guessed baseline, then endless back-and-forth on top of it.

Raymond: And within the same time slice, there might be three Xiaohongshu old-share trades executing today — ten million, twenty million, thirty million — at three completely different prices.

Zhuang Minghao: Genuinely possible. Because background, deal method, named or anonymous, how many structural layers, whether it goes directly on book, payment terms, RMB or USD, individual or institution, identity — all carry requirements. This market has sophisticated buyers too: tell me the format you need and I’ll prepare the matching instruments; different formats, different quotes.

Raymond: Two things I can add. Old-share sales come in two kinds. First: companies with clear ongoing fundraising needs — OpenAI, Anthropic — where old shares typically price at a discount benchmarked to the new round’s pre-money valuation.

Zhuang Minghao: My past experience is always a discount — at least I never backed anything this successful and crazy. And meaningful discounts: 70%, 75% of the round price, I’ve seen both.

Raymond: Second: companies like ByteDance, which hasn’t formally raised from the market since 2018. Beyond bazaar-style haggling, its price moves with one more factor: when the funds that backed ByteDance hit maturity. You watch one waterfall after another — someone at SIG is retiring, a block comes out; some Sequoia fund matures, a block comes out; some angel wants to buy a house, a block comes out. These are genuinely indicators worth tracking.

8. “He holds the floor up for you”

Zhuang Minghao: America’s SpaceX is special, and China’s ByteDance is even more so. In my dataset, built on 36Kr’s message board — about 180-plus issues over five years — ByteDance appears 500-plus times, in almost every issue, on both the buy side and the sell side. It’s a sample with real statistical significance.

Then consider the company’s own posture. First, it’s one of the few Chinese companies that publishes an employee share-buyback price every year — its own mark, adjusted for business and competition. Second, at the first layer — true cap-table equity transactions — if an investor wants out, ByteDance is willing to buy, because ByteDance’s own cash suffices.

With its own mark and its own liquidity, it takes on the role of a yardstick. He holds the floor up for you: whatever happens, my trade will clear. So whether it lists or not makes far less difference to its shareholders. It’s simply too special — you’d struggle to find another company like it.

Raymond: Alright, I’d rather not keep discussing ByteDance — I’ve been second-guessing what I can ask and say this whole time, and the psychological pressure is real. Let’s talk about 36Kr.

9. “From information to a closed deal is a very long way”

Raymond: Your whole analysis rests on one source: 36Kr’s message board. 36Kr matters in Chinese tech media, but it’s hardly the whole market — it’s a sample. I have similar sources abroad: Telegram channels that push daily — selling Databricks today, at this price, this structure, all spelled out. Lately Harvey and OpenEvidence have both had sellers. Foreign FAs (editor’s note: financial advisors who broker deals) come to me asking: hey Raymond, you’re Chinese — got any ByteDance? This information is global. Why were you comfortable building on 36Kr’s data?

Zhuang Minghao: When they launched the column in ‘21, I actually had a serious conversation with 36Kr’s CEO about it. After the first issue I was worried: this is non-public information, and it could hurt a lot of companies. Imagine you’re a company — ByteDance-sized or barely at a 100-million-RMB valuation — and you suddenly find a batch of your shares circulating far below your last round’s price, with no idea who’s selling or why that price. How would you feel?

And 36Kr had, in earlier years, done FA work, incubation — once wanted to build a trading platform. But today it’s purely a message board. Pure display.

Raymond: Do I pay to post?

Zhuang Minghao: No.

Raymond: So could I post: I, Raymond of Mossfire, wish to buy one billion dollars of ByteDance at a $300 billion valuation — would I be beaten to death? He’d just block me, right?

Zhuang Minghao: He might list it. Might adjust it for you against the prevailing mark. But to flip it around: early on people were worried — those from the primary world especially didn’t want this surfaced above water. Yet they’ve kept at it, uninterrupted, for five-plus years, through multiple column editors. After my episode aired, the column’s first editor actually commented under it saying he never imagined someone would write this up — he’s been at a VC for years now.

Why I’m willing to use the data: any single issue surely contains outliers — as you said, anyone can in theory post. But it has persisted five years, never became a platform, never extracted major profit from the endeavor. That time span, plus 180-plus issues and 5,000-plus entries, theoretically supports statistical judgment — the long cycle washes most outliers away. And the trends it yields square with what we remember happening: the industry’s rotation from consumer internet to web3 to hard tech to rockets and robots, the supply-demand of assets — it all matches. It earns its keep, so it’s worth discussing. And with statistics like these, match or mismatch are both useful: a match validates your judgment and firms up a felt trend; a mismatch hands you a surprise.

Raymond: 5,000-plus entries over 5 years — a thousand-plus a year, three a day, four or five per working day. But my impression — insiders in the comments, please correct me — is that they never did FA work, never took a cut in the middle. It’s a bulletin board.

Zhuang Minghao: They considered it, but no. My view stands: this is not an easy business. Information is merely the first link in a trade — possibly the lowest-weighted link in the entire chain. FA fees hinge on closing. But from information to closing — especially in old-share deals, which aren’t financing deals; financing is far simpler — matching buyer and seller, assembling the toolkit, shuttling information back and forth, managing the on-off switches with founders and existing shareholders, above water and below — none of it is easy. Many Chinese tech media have attempted FA; none seems to have done it particularly well, and this is why: from displaying and listing information to an actually closed deal is a very, very long way.

Raymond: Let me add personal experience. I’ve participated in a few of these trades. At the time I never saw the market at your altitude; my understanding was simply: there’s a company, there’s a chance, let’s pool with friends and have a go. Say we want to buy a million — he puts in 300k, I put in 500k. In hindsight, that was the secondary-private market; we bought someone’s stake, we just didn’t call it that. The process is riddled with problems — legal structures, and the problems born of trust; extremely non-standard. You keep herding group chats and wrangling lawyers, and after long enough you say: unless I’m buying truly big, what is the point? The labor is enormous and the payoff wildly uncontrollable — possibly nothing at all.

Let me interject here: have you heard any truly ugly cases? Someone thinks they bought SHEIN old shares, pays 10 million, and three years later discovers the shares don’t exist?

Zhuang Minghao: Not firsthand, but such things certainly happen — I have friends who do exactly these trades. It comes back to the same question: what exactly did you buy? This market exists precisely because buying equity directly is unworkable, so the overwhelming majority of what trades is not direct equity. SPV nesting is the benign case; the Chinese market has more convoluted forms — nominee holding, which splits into individual nominees and corporate nominees — layered on top of the USD-versus-RMB problem. If any single hook in the chain snaps, the whole chain snaps. The more complex your chain, the higher the odds it breaks. I haven’t heard of an especially large default or repudiation — but they certainly exist; we just don’t hear about them.

This circle in China isn’t that large. The FAs and middlemen who genuinely specialize in these trades aren’t numerous; everyone knows who they are. They work at it all year and might close a handful of deals.

Raymond: Like luxury-property agents. And I keep noticing friends’ investment banks playing similar roles — which is why everyone converges on the big banks in the end, all of it collapsing to the American five. My situation now: sometimes I express interest in a company, wanting secondary-private shares, and there’s a post on Xiaohongshu — the moment you so much as like it, someone chases you into your DMs.

Zhuang Minghao: Oh yes. There was that joke — in a Hangzhou KTV you can buy allocation in a certain top company. Honestly, we should thank the recent DeepSeek fundraising headlines—

Raymond: Let me tell you — it wasn’t DeepSeek.

Zhuang Minghao: Right, I know. But those headlines mass-educated everyone on the concepts — much of what we discussed today, people only learned from the news that such games were even possible.

10. “Every individual was rational; the collective decision failed”

Raymond: I want to single out one successful Chinese company: SHEIN. In your grand table it was once glorious and now faces difficulty. On recording day, its listing is four days away.

Zhuang Minghao: Very representative, I think. At peak meaning, $98 billion; the IPO offering is around 26 — how much it pops, we don’t know; it hasn’t priced. Looking back across SHEIN’s arc, valuation is a result: gather every factor into one place and compress it into a single one-dimensional outcome — that’s valuation. At the height of its glory, $100 billion, of course it sought a listing. But the path hit something bigger — problems that weren’t SHEIN’s problems but China-US trade’s problems. So SHEIN once considered becoming a Singaporean company, once considered London, Hong Kong — and originally, of course, America. Looking back down that road: did SHEIN lack lawyers? Lack good investors? Absolutely not. Everyone had the most professional people doing the most professional work.

Raymond: Hold on — I’m not sure I agree with that last line. I’m genuinely curious, with no intent to smear anyone: how did it end up like this? This is a company with rock-solid operations. I used to sell a lot of clothing; I know this business intimately, I’ve toured their factories, I’ve seen that quick-turnaround supply chain many times. They are formidable — the full stack is excellent. In Chinese supply-chain management, truly, SHEIN has no equal. So how did the capital-markets maneuvering end up like this? Did nobody intervene? Do the investors not intervene? Or is it the founder? Or do investors simply have zero influence over him?

Zhuang Minghao: There’s influence, I think. But the problem is: the final call is still his — investors can’t make it for him. Along the way there was positive and negative feedback — the executives he recruited, the money; plenty of negative feedback already. Including the London listing — I don’t even know whose idea that was. All manner of accidental factors stacked into a collective outcome: every individual looks rational — investors, bankers, founder, each gave quote-unquote “rational” advice and feedback — but when it reached collective decision, something failed. A decision permits only one choice, and the world-line we observe is the one that unfolded. There’s luck in it too — much of it beyond his control.

One more view: this company is also past its hungriest, most fearless, most swaggering phase. Choosing this path today may simply be a result of its stage in life. Nothing more.

Raymond: I like this company enormously — deep respect. SHEIN is a singular company, formidable, with no global equal to be found. That these past few years turned out like this, I find genuinely regrettable.

Zhuang Minghao: Which is why listing remains a matter with plenty of metaphysics in it. Too many uncontrollables — timing, external environment, capital markets. It takes heaven’s timing, earth’s advantage, and human harmony, and it truly resists clean explanation by rationality or so-called long-termism.

The state of the secondary-private market and the rotation of SHEIN’s valuation match up: most SHEIN postings were sells, not buys, because the valuation had once been so high and funds had hit their exit windows. But conversely, any company that makes this board at all is a company its holders consider worth bringing to the table — appearing at all means it once had its shining moment. Tuhu appeared many times, but never under its own name — always as “a certain automotive aftermarket company”; only one name in that sector merits discussion, so everyone knows at a glance. Then there’s “a certain autonomous-driving company” — which one? No idea. Could be the one that just listed starting with M, could be one listed for years. The overwhelming majority of postings live in exactly this form.

Raymond: Your “five most sought-after companies”: ByteDance, SpaceX, SHEIN, Neuralink, Xiaohongshu. In ‘22 and ‘23 SHEIN utterly dominated that board, nearly at ByteDance’s level. Rather poignant.

11. “The world before Didi and the world after are two different worlds”

Raymond: Walk us through how the buyer-seller balance shifted across these five years.

Zhuang Minghao: First, the starting point matters. The dataset begins in June 2021 — June ‘21, the Didi incident. I’ve said it countless times: the world before the Didi incident and the world today are two different worlds. After Didi, US listings vanished for years, Hong Kong was weak, and the domestic A-share pipeline was clogged shut. The direct consequence: everyone needed an exit. Recall 2021 through 2023 — remember all the “VC is dead” talk? From early ‘22 to late ‘23, two straight years pointing down. Naturally, in that window everyone sought exits by any means necessary. That was the first great quote-unquote “boost” to the secondary-private market — which is why that era was overwhelmingly sell-side postings, with no buyers.

But the darkest hour turns. As Hong Kong reopened, as A-shares offered hard tech their quote-unquote “support,” and as America loosened somewhat — pure internet companies may be finished there, but less sensitive companies can still list in the US today — buyers multiplied. And what people want to buy tracks the industry’s rotation exactly. What was being sold before was the previous generation: consumer. Heytea appeared many times — peak 60 billion RMB, the byword for consumer startups at the consumer craze’s height. What’s it worth now? Nobody knows. Its latest entry in the spreadsheet is years old — no trades for years. How would you even price one now?

So watch the sequence: first internet platforms and consumer companies, then a wave of web3 and metaverse, then hard tech, then AI, and today robots, commercial space, rockets. Once the rotation arrived, in the new names nobody sells — everyone knows something big is coming — but plenty want to buy. Same story as America: I want OpenAI there, so here, today, I want Unitree (editor’s note: the humanoid-robot company).

Raymond: One question. In ‘21, 87% of postings were sells; this year 57% are buys — the market has genuinely turned. But we can’t confirm those sell-side leads ever closed. It may be one giant bulletin board: back then everyone wrote “I want out” and nothing closed; today everyone says “I want cheap Unitree” and nobody can buy it. It may simply be the mood of the secondary-private world made visible. A vibe.

Zhuang Minghao: Right — vibe trading. As a read on sentiment it’s fine, but it cannot verify the trades themselves — closings are unverifiable, full stop.

Raymond: Then at minimum we know this: the data is large enough, the span long enough, and it matches the assets we actually encounter in our investing — it can serve as a leading indicator of market sentiment.

Zhuang Minghao: Because the most sensitive people always move first. Though nowadays, with AI and modern information flow, transmission time has collapsed. Same as today’s public markets: a company sprouts a new narrative, and from the innermost traders to research analysts to the big trading houses to retail to the media — a chain that once took ages now completes in 24 hours.

12. “321 bosses willing to clock out”

Raymond: The traditional imagination of this market is: I buy ByteDance shares, wait for the IPO someday, sell. But you mention that among these five thousand postings, many are seeking acquisition. What’s that about?

Zhuang Minghao: The poster may not be a pure individual or fund — it may be a strategic player, an A-share listed company, even an A-share company’s affiliate with M&A demand. And you must understand: M&A in China is hard. China has no platform for matching acquisition targets — none. M&A runs on fate. The old way was FAs or banks hunting for you. Now, with the message board, an acquirer can come looking too: direction decided, rough size decided, scan the board for what investment institutions or related parties happen to hold — the company itself may not want to sell, but its investors may want out precisely this way. The share is small, single digits, but it has begun.

Raymond: You noted that in ‘24 acquisition-seeking postings ran high, then collapsed in ‘25. Why?

Zhuang Minghao: Because a lot of them could suddenly list in Hong Kong.

Raymond: So in ‘24, with IPOs looking hopeless, everyone rushed to sell — a for-sale board: buy me, please, buy me. And quite possibly investor-initiated. Among those 321 for-sale notices, did you find any interesting companies?

Zhuang Minghao: I haven’t looked closely. You’ve just reminded me — worth digging back into; there should be some decent companies in there.

Raymond: Here’s my thought: eighteen months ago, before capital markets recovered, these founders already wanted to sell — whether the companies are worth buying we can judge later. In my experience the hardest thing about M&A in China is that the boss is not a professional manager — China is all first-generation founders, and they never clock out. Has Jack Ma clocked out? Certainly not. The nightmare is a company with genuine acquisition potential whose boss won’t clock out. So a lead list of 321 bosses willing to clock out is, I think, extremely valuable. You could clean that up and sell it to this Boss Zhang and that Boss Liu.

Zhuang Minghao: Ha — we could. I hadn’t thought of that. There’s real money in it.

13. “How good would US-China relations have to get before ByteDance lists?”

Raymond: Last question. Having read 5,000 leads across five years of this market’s rise and fall — your deck calls it a leading indicator of sector heat — at this moment, is there any track or any company that could be the next ByteDance?

Zhuang Minghao: Very hard now, I think. ByteDance is too singular — too large, and persistently there. SpaceX has listed; OpenAI and Anthropic will shortly; the American market will keep feeding new names into the real-Mag7 index. But China’s ByteDance will not list in the short term. Run the statistics again in five years and ByteDance is still #1. Quite likely.

Raymond: Wait. I understand that if ByteDance stays private, its size and liquidity make it the deepest market — plenty of supply, plenty of buyers. But why wouldn’t it list?

Zhuang Minghao: My old joke: how good would US-China relations have to get before ByteDance lists?

Raymond: Why does its IPO hinge on US-China relations? Hong Kong, surely?

Zhuang Minghao: I doubt Hong Kong could absorb it. Or rather — the matter is too complex; the considerations at play are beyond what investors like us can even enumerate. Too much contingency. Though possibly, some day, suddenly, it happens. Who knows.

Raymond: My view differs a bit. ByteDance is market-proof — it can list in any market, bull or bear. As long as Hong Kong remains an open market, ByteDance lists there and global investors come running. It would be a company that changes Hong Kong — ByteDance is bigger than Hong Kong. So whether Hong Kong can hold it is not the question; ByteDance would carry Hong Kong up. Put differently: wherever ByteDance goes, it lifts the venue.

Zhuang Minghao: True — ByteDance is bigger than Hong Kong. And in today’s geopolitics, if it ever lists, Hong Kong is the only venue. In America it wouldn’t be so special anyway.

Raymond: Also, I’m marginally more optimistic here, because our AI war needs a great deal of money. A listed company simply has more weapons — wasn’t there a rumor these past few days about ByteDance and the STAR Market? When you know the CapEx is heavy — cards to buy, data centers to build, the whole steel-and-concrete leviathan — a venue where you can borrow, issue a CB (convertible bond), is obviously good. Private companies just have it hard.

Zhuang Minghao: But I suspect Zhang Yiming may feel: I have enough. If I wanted to, I could.

If you're working on this too — or you think we've got it wrong — write to us at [email protected]; if you'd rather not write, just leave your email below.