Summary

China’s 1.5 market was forced into existence by the fund cycle: VC funds have to settle at maturity, but “the exit environment hasn’t given everyone room to settle on time.” The result is an in-between market where buyers and sellers privately agree on a price—neither a financing valuation nor a public-market valuation. ByteDance was founded 14 years ago, and “the angel investors are already due for their second exit,” helping force the rare practice of ByteDance, Xiaohongshu and Kimi buying back employee stock options annually.
SpaceX was originally an asset that required VCs to commit tens of millions to more than $100M, but platforms and SPVs eventually broke the ticket size down to $5,000, with retail investors caring more about Musk’s narrative and “the pie in the sky” than financials. Raymond added that, stripped of the Web3 hype, Web3 “purely from a financial-infrastructure perspective” has genuinely helped. USD capital with Chinese ties may reach SpaceX and similar assets through layers of SPVs; Neuralink may even rank in the top five by trading heat. “Whether what you’re touching is the second layer or the seventh, we don’t know,” but once a trade exists, in theory one can say, “I’m already an investor in this company.”
Raymond’s other “Magnificent Seven” are SpaceX, OpenAI, xAI, Anduril, Databricks, Anthropic and Stripe; SpaceX is described on the show as already public at a $1.8T valuation, while OpenAI and Anthropic will probably list above $1T, bringing the U.S. market its “Fourteen Titans.” Zhuang Minghao summarized the basket as the consensus around AI, defense and chips; like the S&P 500, it only includes companies in the hot sectors, so it rises quickly. Stripe has just acquired OpenRouter and ranks near the top by trading activity.
Discussing private-market pricing, Zhuang Minghao said, “the overwhelming majority is priced on formal,” rather than on a full read of fundamentals. The secondary-share SPV for Zhipu used 2024 data even before its listing—there was “basically nothing to look at”—yet its valuation had already reached roughly RMB16B-18B. “When the rocket is about to launch, don’t worry about your seat—get on first.” By contrast, assets such as Canva and Discord are no longer surging, but have information depth approaching a prospectus, so investors have to fall back on conventional judgment.
ByteDance stands alone: it appeared more than 500 times across 180-plus 36Kr bulletin-board editions, publishes an employee buyback price every year that serves as an automatic valuation anchor, and is willing to use its own cash to take the first layer of secondary shares—“whether it lists or not makes less difference to shareholders.” The two disagree on whether ByteDance can list. Zhuang asks, “How good do U.S.-China relations have to be before ByteDance can go public?” and says Hong Kong may struggle to absorb it. Raymond counters that ByteDance is a market-proven company: “It’s not about whether Hong Kong can handle it; ByteDance would lift Hong Kong,” while its AI capex needs the financing tools available to public companies.
Five years of data show the market’s hot sectors rotating: Zhuang sees the Didi episode as the watershed, while Raymond says the market has changed if 87% of information was sell-side in 2021 and 57% is buy-side this year. The focus has moved from internet and consumer names to Web3, hard tech, AI, robotics and commercial space. Heytea peaked at roughly RMB60B and now “hasn’t traded for several years,” while newer names such as Unitree have no sellers even as many investors want to buy. Both stress that this is only a sentiment bulletin board; in the extreme, “5,000 listings could produce zero deals.”
SHEIN’s decline from a peak valuation of roughly $98B to a reportedly $26B IPO valuation shows how individually rational decisions can fail at the group level and compound with external factors. Cainiao, meanwhile, “was a club deal from day one”: its organizer later ran into other problems and could no longer coordinate or backstop the deal, leaving holders to find their own solution. Raymond sees the 321 acquisition-seeking listings from 2024 as a potentially valuable roster of “owners willing to get off work”—Chinese entrepreneurs rarely do—because “this one could make a lot of money.”

Deep dive

1. The 1.5 Market: An In-Between Market Forced by Fund Maturities

Zhuang Minghao’s primer: the 1.5 market is literally “1.5”—the company is still private, but shareholders want out. Buyers and sellers transact privately, creating “a third kind of price: neither a financing valuation nor a public-market valuation.” The root cause is a mismatch: funds have finite lives and must settle at maturity, “but the exit environment hasn’t given everyone room to settle on time.” Volatility in both China and the U.S. has pushed this once-submerged, highly volatile market toward greater formalization, more regulatory attention and a more visible place in the public conversation.
ByteDance is the extreme case. Founded in 2012 and now 14 years old, “the angel investors are already due for their second exit.” That has helped create the practice of private companies buying back employee options annually; the best-known examples today may be ByteDance, Xiaohongshu and Kimi.

2. SpaceX: From a Tens-of-Millions Entry Ticket to $5,000

Zhuang’s framework is that SpaceX combines Musk’s record of success with a valuation that keeps climbing. Retail investors care less about revenue, profit and gross margin than about the story Musk is building and “the pie in the sky.” Under the old system, ordinary investors either became LPs in a top-tier fund and faced a very high threshold, or stayed out altogether. Platforms and some so-called transaction structures broke the ticket into smaller pieces, potentially down to $5,000.
Raymond added the American idea of “financial democratization”: why should the richest people be able to buy the best companies and get even richer while ordinary people are shut out? He had previously discussed a platform that trades assets in tokenized form; after SpaceX’s listing news emerged, the platform’s price doubled.
Stripped of the Web3 hype, Raymond believes Web3 “purely from a financial-infrastructure perspective” has genuinely helped. Zhuang agreed that the era, the company and the individual all have to be in place. Both retained reservations about legal deficiencies in the chain and the fact that the connections are not fully established.

3. How Chinese Money Gets Into SpaceX and Neuralink

The surprise in the data was Neuralink’s appearance in the rankings, potentially placing it in the top five—ahead of Discord, Cainiao and DJI. Zhuang’s explanation is that USD capital with Chinese ties may come from co-founders after an internet company’s IPO or from retired investment professionals. They need assets but may lack the capacity to set up a family office or become LPs in a top-tier USD fund.
Individuals have difficulty buying equity directly, so they participate through nested SPVs. As Zhuang put it, “whether what he’s touching is the second layer or the seventh, we don’t know.” The account is denominated in dollars, but it is difficult to know who that dollar ultimately came from. Once a trade exists, in theory one can say, “I’m already an investor in this company.”
Raymond took the joke further: put $100 into the trade, with $50 going to OpenAI and $50 to Anthropic, and you can claim to have invested in both companies. Both stressed that this was not investment advice; the chain contains numerous legal gray areas and links that are not fully connected.

4. Robinhood’s Tokenization: Buying an IOU and Trust

Robinhood launched pre-IPO shares of OpenAI and Anthropic in the second half of last year, after which both companies said the arrangement was illegal and unrecognized. Raymond summarized Robinhood’s possible logic as: “The stock layer is a stock matter; the token I attach here is a token matter.”
Zhuang’s breakdown is that the buyer is actually “buying a certificate, an IOU, from Robinhood.” If the company eventually lists, the price can be mapped to the underlying shares. But the buyer does not own that portion of the equity; the buyer is trusting Robinhood not to disappear, create problems or refuse to honor the claim.
Even though Robinhood is a large, public trading platform, Raymond still believes the arrangement may be problematic under a strict reading of the law.

5. People Buy in the Restricted Zone—and Some Try to Delete the Story

Raymond said that under the current U.S. regulatory framework, Chinese people cannot invest in U.S. AI companies. In the first half of 2026, during the round that valued OpenAI at about $380B, he saw many briefs and documents asking specifically about the investor’s Chinese identity.
Zhuang shared a piece of gossip: a Web3-oriented media outlet used a cap table to describe how China-linked funds participated in SpaceX through SPVs. After publication, many people contacted Zhuang seeking an introduction to the journalists so the article could be deleted. The money involved may have been only $20M or $30M invested by some funds years earlier, but the outlet named specific funds. The names may have been written only in pinyin, yet “the people who know will know who it is.”

6. Another “Magnificent Seven” and the Trillion-Dollar Club

Raymond listed another “Magnificent Seven” on his presentation: SpaceX, OpenAI, xAI, Anduril, Databricks, Anthropic and Stripe. SpaceX is described on the show as already public at a $1.8T valuation. Raymond believes OpenAI and Anthropic are about to list and will probably debut above $1T, taking the U.S. market from the “Magnificent Seven” to the “Fourteen Titans.”
Zhuang described the group as the consensus in the U.S. primary market over the past year or two: AI, defense and chips. Money, talent, attention and the possibility of an IPO have all concentrated there, while revenue, users and valuations are growing rapidly. The group resembles the S&P 500: it includes only companies in the hot sectors, so the index appears to rise quickly. Companies such as Canva that are no longer surging may not make the cut.
Stripe sits at the bottom of the list. Zhuang said it had just acquired OpenRouter, and its conference could now be compared with events held by AWS, Google Cloud and even Nvidia. It is seen as AI payments infrastructure; with no financing for an extended period and no clear IPO news, demand for its secondary shares is active and its ranking is correspondingly high.

7. Sam Altman and Employee Secondary Shares

One interpretation from Zhuang is that Sam Altman helps OpenAI employees sell secondary shares every year, which may be his most important job of the year. Raymond added that employees sold to SoftBank last year, while this year’s transaction should already be complete, at a price in the $70-plus range. Executing a large, high-valuation transaction while distributing proceeds compliantly is not something just anyone can do.
Raymond also said Anthropic conducted an employee stock tender or employee-equity tender in April or May this year, although he does not know exactly with whom Anthropic completed the transaction.

8. Private-Market Pricing with a Heavy “Formal” Weight

Raymond used PDD as a comparison. In public markets, investors can review the past 8 quarters and listen to earnings calls, so information is transparent. With Anthropic, it is still unclear whether ARR is net or gross, how much revenue is recurring, and whether users churn after token maxing. He asked how much weight private-market pricing assigns to formal versus fundamentals.
Zhuang’s direct answer was: “The overwhelming majority is given to formal.” In Zhipu’s case, the pre-IPO secondary-share SPV still used 2024 data—effectively, there was “basically nothing to look at”—but the valuation had already reached roughly RMB16B-18B. An intermediary can prepare a BP, documents and forms, but the materials may be stale or merely public information. The decision ultimately becomes: “Do you bet, or do you not bet?”
Zhuang described rapidly rising assets with the line, “When the rocket is about to launch, don’t worry about your seat—get on first.” By contrast, companies such as Canva and Discord that are no longer rising sharply, or whose fundamentals have changed little for a long time, may have information depth approaching a prospectus. Investing in them is therefore closer to a conventional fundamental judgment.

9. Zombie Unicorns and Used-Phone Recycling

Raymond reviewed the unicorn rankings from Hurun and Ren Zeping and found that many names outside the top 10 had been on the list since 2015 and were still there 11 years later. Zhuang noted that IT Juzi had published a “disappeared unicorns” list as early as 2015.
These companies once looked great in LP letters and IRR calculations, but become difficult to handle when a fund reaches maturity and has to settle. An S fund or secondary fund may sometimes buy the entire fund interest, using original cost as the valuation basis.
A fund may hold 10 portfolio companies, each with a different condition, cost basis and reasonable valuation. The usual process is to audit the portfolio as a whole, then determine a transaction price. Most assets should be written off; only a small number with enough value to discuss move on to debates over a discount, par or premium.
Raymond compared the process with used-phone recycling: an iPhone bought for RMB6,000 may be taken back for RMB1,000, with its memory or other parts removed and sold while the rest is discarded. If the condition is good enough, it may simply be resold. Zhuang said every outcome is possible.

10. Xiaohongshu: Different Prices Can Exist at the Same Time

Xiaohongshu first appeared on the market at a valuation of about $12B. Its last verifiable price was probably $38B, above the $32B relayed by Zhu Xiaohu. More recently, with no new information, the price in the round bought by DST may have been in the $20B-plus range.
Zhuang said the underlying business remains the foundation, followed by trading sentiment, the IPO timetable, U.S.-China external factors and the ticket sizes available to buyers and sellers at a given moment. If several funds mature around the same time and short-term selling pressure becomes concentrated, the price will naturally be weak.
The market is highly non-standardized. At the same point in time, there could be three small Xiaohongshu secondary trades at RMB10M, RMB20M and RMB30M, with completely different prices. The reasons may include the transaction background, the number of structural layers, whether the buyer enters the cap table directly, whether the deal is in RMB or USD, whether the buyer is an individual or institution, payment terms and investor-identity requirements.
Sophisticated buyers prepare different tools for different transaction structures, then quote different prices based on the structure.

11. Why ByteDance Stands Alone: Self-Pricing and Self-Buying

ByteDance appeared more than 500 times across more than 180 editions of the 36Kr bulletin board, almost every edition, with both buy and sell listings. Raymond believes ByteDance’s annual employee buyback price provides an automatic valuation anchor. If secondary shareholders who are actually on the cap table want out, ByteDance is also willing to buy, because it has enough cash to act as its own liquidity pool.
Raymond distinguished between two secondary-share pricing systems. Companies with a clear financing need are usually priced off the pre-money valuation of a large new round, with secondary shares potentially trading at a discount—historically, 30% or 25% discounts have been seen. Since ByteDance has had no formal market financing since 2018, its price is driven more by secondary trading and fund-maturity selling pressure.
Someone at SIG retiring, a Sequoia fund reaching maturity or an angel investor needing to buy a home can all create supply. ByteDance has its own employee buyback price and some liquidity, so for holders, whether it lists makes less difference.

12. The 36Kr Bulletin Board: The Statistical Value of 5,000-Plus Listings Over 5 Years

The data comes from 36Kr’s database and bulletin board. The series began in June 2021 and has run for more than 5 years, covering more than 180 editions and over 5,000 listings. When Zhuang launched the column, he had a serious conversation with the 36Kr CEO and worried that exposing secondary shares trading below a financing valuation could hurt companies, especially because neither the seller nor the reason for the price was transparent.
36Kr previously worked as an FA, incubator and attempted transaction platform. Zhuang now defines the product as a bulletin board that only displays information and does not broker deals. In theory, anyone can post an unrealistic buy or sell listing, so every edition will contain outliers.
Zhuang still uses it because the column has lasted 5 years, generated no genuinely massive profits on the platform and accumulated a long enough cycle and more than 5,000 data points. Raymond believes a long time series can filter outliers to some degree, while the sector rotations visible in the data broadly match real-world experience.
Raymond said overseas Telegram groups and channels post Databricks quotes every day, and there are also sellers for Harvey, Evidence and Anthropic. Overseas FAs have even asked him whether he has ByteDance shares.
Raymond also shared his own AI experiment: he compiled more than 20,000 LinkedIn résumés from OpenAI, Anthropic and DeepMind. More than 1,000, nearly 2,000, of the names were in Chinese pinyin, and more than 300 had a clear connection to China. The idea was inspired by an article titled “Who Is Making Binance?” Without AI, the work would have been “absolutely impossible.”

13. Information Is Far from a Deal: FA Is Hard and Chains Break

Zhuang’s view is that information is only a low-weight component of a complex process. Secondary transactions are far more complicated than financing deals: they require information transfer, communication with founders, company shareholders and existing investors, and the handling of onshore, offshore and other structures. Displaying information is a long way from closing a deal, which is why no Chinese tech-media company has emerged as a particularly effective FA.
Most transactions do not involve buying equity directly. They run through SPVs, nominee holders who are individuals, nominee companies or nominee funds, with USD and RMB structures layered on top. If a single link breaks, the transaction chain breaks; the more complicated the chain, the higher the theoretical risk of failure.
Zhuang has not personally heard of any especially large or obvious cases of a seller walking away or refusing to honor a deal, but believes they certainly exist and are simply unknown to the outside world. There are not many FAs and intermediaries specializing in these trades; some may close only a handful of deals a year. Zhuang compared them with luxury-home brokers.
Raymond shared an experience of pooling money with a friend to buy secondary shares: the friend contributed RMB300K and he contributed RMB500K. Because the transaction was so non-standardized, the process involved constant group chats, lawyers and legal structures. The labor required and the payoff were highly unpredictable; the outcome could even be nothing.
Raymond added that DeepSeek financing coverage helped popularize these mechanics. Today, liking a Xiaohongshu post can prompt someone to DM an offer for a stake, while jokes have circulated that shares in a leading company can be bought in a karaoke club in Hangzhou.

14. SHEIN: Strong Operations, Uncertain Capital Markets

Raymond said SHEIN’s valuation approached $98B at its peak but was reportedly about $26B in its IPO prospectus, and it was still waiting 4 days before the listing. He considers SHEIN’s business and rapid-response supply chain extremely strong; after visiting its factories several times, he said it would be difficult to find a Chinese company with better supply-chain management.
Raymond believes SHEIN’s capital-markets problems were not necessarily all self-inflicted. They were also shaped by U.S.-China trade relations and other external factors. The company considered incorporation in Singapore and listings in the U.K. and Hong Kong, with the U.S. as the earliest option.
Zhuang said investors obviously have influence, but the final decision belongs to the founder; investors cannot make that decision for the founder. Investors, bankers, founders and executives may each offer feedback that appears rational, yet the group decision can still fail. Timing, external conditions, capital markets and luck then compound the outcome.
Zhuang does not know whose idea a London listing was, but believes SHEIN may have passed its most fearless and swaggering phase. Raymond still expressed respect for the company and regret over how it has changed over the years.

15. Cainiao: A Club Deal from Day One

Raymond sees Cainiao as one form of a Chinese club deal: participants want the person who organized the club to handle overall coordination and provide the backstop, but that organizer ran into other problems over the past year and could no longer carry the burden. Cainiao is also relatively peripheral to the Alibaba system.
Zhuang’s judgment was more direct: “Cainiao was a club deal from the first day, so no one will take it. Who would?” The show used a dinner-table analogy: everyone wants to come eat at your house, but the cook has gone off shift, so all that remains is ordering takeout and making do.

16. Five-Year Rotation: From a Predominantly Sell-Side Market to More Buyers

The data begins in June 2021, just as the Didi episode unfolded. Zhuang believes “the world before Didi went wrong and the world today are two different worlds.” U.S. IPOs effectively disappeared, Hong Kong equities languished, and the A-share market looked clogged, as if the gates had closed. Everyone needed an exit, and “VC is dead” became a common refrain.
As Hong Kong gradually reopened, A-shares began supporting hard tech and the U.S. market showed signs of loosening, the buyer base expanded. The sectors rotated from internet platforms and consumer names to Web3, the metaverse, hard tech and AI, then to robotics and commercial rockets. Heytea may have reached a peak valuation of RMB60B, but later “didn’t trade for several years.” Newer names such as Unitree attract many would-be buyers but typically have no sellers.
Raymond’s example was that if 87% of listings were sell-side information in 2021 and 57% are buy-side this year, the market has reversed. He also cautioned that intentions posted on the bulletin board do not necessarily result in completed trades.
That is why both view the board as a sentiment bulletin board and a leading indicator of market mood, not a record of completed transactions. In the most extreme case, “36Kr could publish 5,000 advertisements and not a single deal would close.” It helps gauge sentiment but cannot solve the transaction itself.

17. 321 “Sell-the-Company Notices” and the Next ByteDance

Acquisition-seeking listings accounted for a relatively high share in 2024. The potential buyers included strategics, A-share listed companies and their affiliates. China has no unified M&A platform, so deals often rely on FAs, investment banks or serendipity to find targets. These listings declined in 2025, and Zhuang agreed that more companies may have gained the ability to list in the U.S. or Hong Kong.
Among the 321 “sell-the-company notices,” the initiator may have been an investor or a founder, but in many cases it was an investor. Raymond believes companies willing to sell 18 months ago, before capital markets had recovered, deserve study. Chinese entrepreneurs generally do not want to get off work—“Has Ma Yun gone home? Of course not”—so the list of owners willing to exit may itself be valuable. Zhuang admitted he had not yet examined it closely but said it could be worth mining.
On “the next ByteDance,” Zhuang said it would be difficult because ByteDance is so unusual. He believes it cannot list in the short term and joked: “How good do U.S.-China relations have to be before ByteDance can go public?” He also thinks Hong Kong may struggle to absorb ByteDance, while leaving open the possibility that “one day it suddenly lists—who knows?”
Raymond explicitly disagreed. He considers ByteDance a market-proven company and believes it can list in Hong Kong as long as Hong Kong remains an open market—perhaps ByteDance would be the company that lifts Hong Kong’s market. He also said the AI war requires enormous capex, and public companies have access to borrowing and convertible-bond issuance, among other financing tools that private companies struggle to use. There have also been rumors of a ByteDance listing on the STAR Market.

Verification Notes

“Formal” was not further explained in the original conversation and is therefore retained as is, rather than equated directly with FOMO. Raymond’s 87% and 57% figures were illustrative examples, and bulletin-board listings do not equal completed transactions.

Why World Models Could Change Robotics, 3D, and Creativity Why AI Agents Could Finally Reinvent the Credit Card