Vol.78 Talking Venture Capital with Lao Qian—A Cross-Podcast Meetup
Summary
The core of venture capital is not a replicable model but “people”—a highly composite variable that庄明浩 ultimately folds into “fate.” Founders matter, but investors’ education, personalities, positive and negative feedback, and path dependence also determine what they choose; industry research, valuation, and terms can support action but rarely constitute sufficient or necessary conditions for success. The bigger the deal—ByteDance, Pinduoduo, Xiaomi, or Kuaishou—the less likely retrospective analysis is to overlap with the true causal chain. “You have to find the thread you can believe in,” but that thread may also be a placebo.
VCs do not make money through a high hit rate; they keep shooting until a tiny number of home runs pull the entire fund upward. A roughly $200M early-stage fund might invest in 50 companies, with an initial check of $3M–$5M typically representing 1%–2% of the fund; only 1–2 investments may determine the ceiling, while another 5–10 keep the fund from losing money. Achieving 2x–3x DPI after 10–12 years is already extremely difficult, because breaking even is meaningless unless LPs are compensated for the long lock-up, illiquidity, and opportunity cost.
Investment committees are naturally good at saying no, but “put a dog there and it will say no 90% of the time” may also be correct; what is truly scarce is taking responsibility for a single yes under uncertainty. When a deal moves from an investment manager to the IC, the essence is to push different decision-makers’ experience, memories of losses, and weights above 60% or two-thirds; regardless of how much research has been done, the final decision is still binary. This was庄明浩’s dilemma with the metaverse when he left Matrix Partners China: passing might have saved money in hindsight, but top funds “do not win wars by saying no and saving money; they win by shooting and betting on something big.”
Fund inflation is directly pushing up company valuations, shortening financing cycles, and sending capital and talent toward the top. A team that once raised RMB200M and wrote checks of several million yuan may still target similar stages after growing its fund to RMB800M–RMB1B, leaving it no choice but to increase both check sizes and valuations; top funds may even fully price in expectations three years out, leaving later entrants no round to join. 庄明浩 cited the US fundraising distribution: the Top 10 funds receive roughly 50% of the money, and the Top 25 about 85%. “Raising the money is only the first step. Money is the ammunition.”
The AI-bubble debate has moved from “is there one?” to “what kind, what happens when it breaks, and should we participate in the final 6–12 months?”—turning the investment question into one of position sizing and odds, not merely direction. 庄明浩’s personal preference may persist for a while, but he believes a fairly large correction is inevitable; his stress test is Nvidia falling 40% from $180–$190, or even quickly returning to the $80 level seen earlier this year. Lao Qian went further, suggesting a 2% blind bet analogous to an early-stage investment, or using puts to seek better odds. The gap remains the same: knowing does not necessarily translate into doing.
The 2021 fundraising cycle determined who could keep investing through the 2022–2023 trough, meaning so-called contrarian investing was often pre-allocated by the capital cycle. Top funds had already raised three years of ammunition before the Didi incident, and themes that later blossomed—semiconductors, humanoid robots, domestic GPUs—had already been claimed. State-backed funds in third- and fourth-tier cities, with RMB5B–RMB6B of AUM, faced three constraints at once: no new money, overpriced hot deals, and no standing to tell the next story. “Whose money you raise, what story you tell, and what you invest in” are not three separate questions but one timeline.
Personal returns for VC professionals arrive more slowly, within a narrower band, and depend heavily on their entry-level position and luck across cycles. Analysts and investment managers generally do not receive the core carry pool; even after making partner, they must invest their own money. 庄明浩 gave the example that a $200M fund requires at least 10% of the GP’s own capital, followed by the fund’s operating period and potentially four years of vesting. A person must first win promotion, then win the fund, the investments, and the exits; by the time they truly enter the “harvest period,” they may be around 50. Delayed gratification is not a cultural slogan—it is the cash-flow structure of the profession.
Judgment about people cannot be quantified; group decisions, diligence, and values-based rules are attempts to improve the odds, not models that eliminate randomness. 庄明浩 admits that “continuous iteration, evolution, and vitality” are empty words if they cannot be measured. A fund can only have different teams, post-investment, finance, and HR leaders repeatedly meet founders, treat each person as a factor, and gradually develop its own taste. He invests in decent, good people rather than “winners who will do anything,” and keeps one empirical rule: if a founder insists that the fund must never invest in a competitor, he says no immediately. That rule comes from three or four cases that were “never right,” but he still refuses to present it as universal truth.
Deep dive
1. “People” is the most important—and most unfalsifiable—answer
庄明浩 gives people the highest weight in early-stage investing because what the industry calls “entrepreneurial spirit” or “founder spirit” ultimately resides in the individual. He immediately acknowledges, however, that this answer contains ability, personality, experience, preferences, and environment, making it inherently larger than any single variable.
An investor’s education, family, upbringing, and the areas in which they have received positive or negative feedback shape how they select people; a founder’s “fate” interacts with the investor’s fit and game. Keep asking why, and investment decisions begin to resemble the complex systems described by fortune-tellers.
The system is dynamic as well: a weighting that works this year may fail next year because time, the cycle, and the competitive environment have changed. “The marginal weight of one factor can be absolutely dominant,” yet professionals struggle to draw that factor clearly while acting.
2. Research supports action but rarely explains supernormal returns
庄明浩 says that when actually doing the work, one is still thinking about specific founders, products, markets, competition, valuations, and terms. If you keep thinking about fate and chaos, you cannot work. Only after 10 or 20 years, having watched countless rises and falls, can people abstract the underlying details into “venture capital is fate.”
His skepticism about retrospectives on ByteDance, Pinduoduo, Xiaomi, Kuaishou, and other landmark deals is that the logic presented by core investors is mostly post hoc synthesis and linear extrapolation. With an omniscient view, “the causal component might not even be one-thousandth.”
Lao Qian initially understood “it is all fate” as “they were simply lucky.” 庄明浩’s correction is that reasons certainly exist, but they may not be sufficient or necessary conditions. Investors still have to find a thread they believe in; otherwise, they cannot pull the trigger amid chaos.
3. Many investment frameworks are merely necessary placebos
庄明浩 calls day-to-day analysis a “placebo,” and says it may not even qualify as true dimensionality reduction. It gives people a sense of security in front of an inenumerable future, but may not determine the final result.
His counterexample is brutal: “Put a dog here to do early-stage investing,” give it only a no button, and it will be right 90% of the time. Even if you give it a yes button and let it press at random, its hit rate may not be lower than that of an experienced investor.
That does not mean detailed work can be skipped. The investor’s job is to be slightly “better” than that 90% mechanical veto: ask questions, enumerate possibilities, and compress the unknowable future into a range of risk they can bear.
4. An IC is a persuasion exercise to push weighted support above two-thirds
A deal reaching the investment committee means the investment manager has already persuaded multiple layers of colleagues that it could generate enormous returns; the founder must likewise face partners and GPs and seek support.
Every decision-maker brings path dependence shaped by age, experience, and the types of projects on which they have made or lost money. The company and its sponsor must eliminate negative buffs and accumulate positive ones, ultimately pushing the weighted vote above 60% or two-thirds.
庄明浩 describes the purpose of the IC in blunt terms: every statement, response, and piece of material exists to push the weighted average above 66%. The research process can be infinitely complex; the capital decision has only two outcomes—invest or do not invest.
5. An early-stage fund makes 50 bets for one or two home runs
Using a roughly $200M dollar fund as an example, 庄明浩 gives a common portfolio of around 50 companies. In the early days, the first check in an A round was usually $3M–$5M, or about 1%–2% of the fund—the first shot.
Of those 50 companies, perhaps only 1–2 are home runs with “no ceiling”; one investment may return several times the entire fund. Another 5–10 may exit through acquisitions, buybacks, or ordinary IPOs and ensure that the fund does not lose money.
Many of the rest may produce no outcome. VC’s ceiling is determined by a tiny number of outliers, so the portfolio’s average accuracy is not the key metric. What matters is whether, while “turning over stones,” you find the one with infinite upside.
6. Texas Hold’em maps rounds, information, and chips with unusual precision
The angel round is like seeing only two hole cards: education, résumé, and rank at ByteDance or Alibaba are all judgments about the cards themselves. A 1%–2% initial position is like posting a blind with almost no information.
By the A round, three community cards are on the table and many things have become visible. Industry, competition, timing, and initial company data begin to emerge, and the investor decides how much to bet, how deeply to participate, or whether to fold. More cards are dealt in later rounds until the hand is revealed at an IPO or exit.
Fund size determines the depth of the chips on the table and whether the investor has room to make further decisions. 庄明浩 believes today’s AI table has become “so distorted that it barely belongs to VC”; many institutions can only choose between zero and all-in, while the entry bar is too high to get seated.
7. 2x–3x DPI is the qualifying line—and extremely hard to sustain
A fund commonly operates for 5–7 years, followed by an extension period; full closure usually takes 10–12 years. If an LP invests 1 yuan and ultimately receives 2–3 yuan, the fund can be called good or qualified.
庄明浩 roughly estimates that this may correspond to an IRR below 10%, yet delivering 2x–3x DPI consistently across multiple vintages remains extremely difficult. “Qualified” is relative to LP asset allocation; it does not mean the business is easy.
Looking at long-term industry returns, he says VC has historically underperformed the S&P 500, while fund outcomes are as extreme as company outcomes: roughly 2% of funds capture more than 95% of industry returns. VC is not a stable compounding vehicle but an allocation to tail possibilities.
8. Leadership has inertia, but every cycle produces new challengers
Sequoia is called the “King of VC” not because it dominates every stage, but because it stayed near the top from the internet era through mobile internet. At the same time, every cycle produces funds that suddenly jump into the highest-return tier, while others fall away.
庄明浩 uses Index as an example. He says Index made roughly 1,000x on Figma and also held the AI-security company Waze, acquired by Google for $32B, as well as investments such as Notion and Roblox.
Funds Index has been investing from in recent years may already have generated more combined returns than Sequoia did over the past 10 or 20 years. “This wave happened to catch it,” remains 庄明浩’s key qualification. Only after exits does the market begin retrospectively explaining why Index succeeded.
9. Fundraising for the third and fourth vintages filters out the face-value funds
A first fund can be raised on personal reputation, relationships, and background. When raising a second fund in year three, investments have usually not exited, but the manager can still show off marquee companies and paper IRR, using face value and an early upward curve to persuade LPs.
By the third fund in year six or seven, the first fund must have generated at least some distributions. By the fourth fund around year 12, the first fund should be finished and the second close to finished. The manager can then rely only on actual results.
庄明浩 summarizes the hurdle for reaching a fourth fund as “winning two and a half times”: at least one home run in each of the first two funds, plus half a home run or a sufficiently clear signal in the third. Hitting consistently across 12 years of technological, policy, and thematic change is therefore extremely difficult.
10. Fund life forces investing and exits onto a single timetable
A fund typically completes most of its initial investments in the first three years while reserving capital for follow-ons. As companies develop over three to five years and costs rise, the manager can continue investing through later Seed, Growth, or parallel funds, matching the fund cycle to the company life cycle.
A fund must settle its books at maturity, so a manager cannot make large new startup investments in year five and assume they will exit within two years. Fund life is not an accounting backdrop; it is a hard constraint from initial investment through follow-on and exit.
Top institutions raise funds of different sizes and stages simultaneously precisely to avoid having one fund carry both early-stage blind bets and large growth investments at maturity.
11. The exit pipeline determines whether the VC business can close the loop
China historically expected LPs to receive cash only, but large A-share sell-downs are complicated and a fund may not be able to sell smoothly before maturity. In-kind distributions were later piloted, allowing funds to return listed-company shares directly to LPs and relieve exit bottlenecks.
庄明浩 warns that this is not a simple transfer. Entry cost, distribution price, tax basis, pricing period, share transfer, and listed-company requirements all interact and require a complete rule set.
“Venture-capital funds make money through exits.” Once exits are blocked, the entire raise-invest-manage-exit chain breaks down. A higher paper valuation does not mean the fund has achieved a commercial result.
12. Looking only at DPI corrects paper prosperity but distorts behavior
An early-stage fund may have DPI of only 0.1 in its first five years because deployment, growth, refinancing, and exits take time. The paper IRR visible in the meantime is “too easy to write”; even with strict audits, there is substantial room for adjustment.
Chinese LPs in recent years have moved from “ignore IRR and look only at DPI,” refusing to hear more stories about marquee companies and asking only how much cash 1 yuan will return over what period. 庄明浩 acknowledges that this focuses on the result but compresses the time needed for the result to appear.
When a fund must prove its diligence at the settlement date, companies that are underperforming, unable to IPO, and unable to buy back shares may be sued. Founders becoming “deadbeats” because of buyback obligations are an extreme example of DPI pressure reshaping industry behavior.
13. The 2022–2023 contrarian opportunity was allocated in 2021
Lao Qian imagines that betting during the industry’s darkest period in 2022–2023 could have produced substantial returns today. 庄明浩’s rebuttal is that this view carries hindsight bias: investment capacity depends on the fundraising cycle that came earlier.
Before the Didi incident in 2021, most funds in the top third had already raised large new vehicles. Even after the macro and entrepreneurial environment deteriorated, top institutions still had enough ammunition; they had been targeting companies expected to emerge around 2023 in the first place.
The semiconductor, humanoid-robot, and domestic-GPU themes discussed today had already been claimed by capital. Saying “we should have bought at the bottom” after seeing them bloom ignores the fact that rounds, access, and capital supply had already changed.
14. Local state-backed funds face a mismatch between the liability side and investment reality
庄明浩 once met the son of the head of a state-backed fund in a third- or fourth-tier city. The fund had total AUM of roughly RMB5B–RMB6B—not small—but the city lacked a distinctive industrial identity like Hangzhou, Chengdu, Wuhan, or Nanjing.
The valuations and chips in hot themes were already fully stacked. If the fund told a newer, more distant story, investors would ask, “Why you and not a top fund?” If it continued telling an existing story, it could only take responsibility at extremely high prices.
The real problem was therefore “whose money to raise, what story to tell, and what to invest in.” All three had to make sense along the fund timeline. The young man was even considering giving up the succession and trading US stocks instead—a sign of how institutional constraints push talent toward more liquid markets.
15. China does not need this many VCs; zombie funds simply have not been deregistered
庄明浩 believes VC should not have received such a high social profile given its capital scale, return stability, and commercial nature. It was elevated into the spotlight because supporting innovation gives it an inherent aura of righteousness.
At the peak, more than 10,000 early-stage private funds were registered in China; the number may later have fallen to 5,000 or 6,000, but he still considers that too many. Many institutions cannot raise a next fund yet must continue managing old funds, auditing, filing, and reporting to LPs, becoming “zombie funds” in all but name.
He cited statistics from around 2021: only 155 publicly identifiable funds deployed capital more than five times during the year. Business-registration, news, and exchange data all miss underwater transactions, but the gap between active institutions and registered institutions is already striking.
16. Bubbles leave ruins but also finance and staff the next cycle
庄明浩 believes 2021 was unquestionably an economic bubble, with many Chinese stocks later falling 80%–90%. But bubbles can also nourish the next wave of opportunity, with AI as one example.
Investors in their 60s and 70s who have lived through multiple cycles are often less extremely pessimistic than younger participants because they have seen curves rise again after deep troughs. Experience cannot predict the bottom, but it changes one’s confidence in the claim that “this time is different forever.”
Age, wealth, target wealth, and an individual’s utility curve also change risk appetite. What looks like “running blind” to an outsider may still be only a blind bet to someone who has already accumulated substantial capital.
17. The AI bubble has already been differentiated four times over
Lao Qian observes that in Q2 2025 people were still asking whether there would be a bubble; by Q3, most were asking whether it was a bubble; by Q4, the bubble was assumed, and the debate had shifted to whether it was benign or non-benign, debt-driven or equity-driven.
The discussion then moved to “what happens even if it is a bubble?” and “what happens even if it breaks?” Markets repeatedly compare AI with the internet, railroads, tulips, Web3, and Chinese real estate. The question has moved far beyond the initial technology judgment.
Lao Qian relays a comment from a Goldman Sachs banker or brokerage executive: the final 6–12 months of a bubble are often the craziest, and the highest returns may arrive during the most dangerous stretch. The question is therefore not whether to identify the bubble but whether to take the “fattest piece.”
Lao Qian compares this with bargaining over old apartments in first-tier cities. Beyond the public listing price, the final cut made in a Lianjia negotiation room is the most brutal; at a bubble top, the reverse is true: the final stretch brings the most aggressive bids and may also deliver the highest returns.
18. Macro affects fundraising, but frontline investors are still required to keep shooting
Startup projects resemble accelerated species selection: within two or three years, an ecosystem completes financing, competition, and淘汰; capital then shifts to the next ecosystem, and even the ruins left by the previous round do not prevent the next one from beginning.
庄明浩 believes macro conditions are felt more directly by fund managers raising capital, while frontline investors at top institutions are less affected. If you work at a fund around China’s Top 30, your job remains to screen from the bottom up, see projects, and shoot. “This is not something you can think about.”
Institutions do not stop operating because of a macro view. Their fund lives and deployment requirements themselves force continued betting.
19. Carry is a longer, narrower bridge than startup options
A fund may take 12 years to finish, after which an individual’s income may remain subject to roughly four years of vesting. The position a person holds when the fund is established already determines how much carry they can participate in.
Analysts and investment managers generally are not in the largest carry pool; they can only climb through promotion and bonuses. If they make partner within three to five years, the next fund still requires the GP to invest personal capital first. 庄明浩 uses the example of investing at least 10% of a $200M fund personally.
庄明浩 cites Peter Thiel’s roughly 65% personal investment in his first fund as an extreme example of conviction. The ordinary young partner’s reality is: “Before seeing any money come back, put money in first,” then wait for the company, fund, and exit to succeed together.
Very few people complete both stages. The first is leveling up through the ranks; the second is leveling up again after committing capital. A small number of professionals around 50 have entered the harvest period and can live relatively comfortably as advisers, figureheads, travelers, or golfers, but that is not the industry norm.
20. Good markets create new funds but lock personal fate into a decade-long cycle
The 2014–2015 “mass entrepreneurship and mass innovation” campaign and the A-share peak created an LP base for new funds. As listed-company bosses and successful founders grew wealthier, they could in turn support young investors setting up independently.
A similar market appeared again in 2021. 庄明浩’s explanation is that running one’s own fund means entering the carry path early and taking the largest share, so whenever capital is abundant, younger investors in the industry emerge to start funds.
But whether a new fund reaches its second or third vintage depends on real exits more than a decade later. A market provides the starting point; disappearing themes, policy changes, and fund life determine the ending.
21. From 2015 to 2025, 庄明浩 moved from participant in the wave into middle age
庄明浩 left VC to work on Panda TV in 2015 at age 29. Over the next decade he experienced a failed startup, a family growing from two to four people, the 2014–2015 and 2021 cycles, and the extreme tests that COVID imposed on life and the industry.
For consumer internet, mobile internet, social, and entertainment, the pandemic pushed the time users could devote to products to an extreme. Time spent, retention, addiction, and business extension were tested almost all at once. It was not normal growth but a test of the boundary.
What he is most grateful for is not one isolated achievement but gradually learning between ages 30 and 40 who he is, what he is good at, and what kind of work suits him. His lower risk appetite was not a sudden pivot; family, failure, and opportunity cost gradually entered the same weighting table.
22. Financial theory can describe power laws but cannot choose for you
Lao Qian mentions Baillie Gifford, the British institution that moved from investing in palms, oil, and value stocks during globalization to growth stocks after 2000, positioning itself as “venture capital in the public markets.”
It used research on increasing marginal returns, bubbles and technological revolutions, and extreme power-law distributions to support moving earlier, eventually entering the semi-private market. 庄明浩’s response is: “There is nothing new.” The interaction of technology, economics, and finance has been studied repeatedly.
He studied technological economics as an undergraduate and graduate student and then spent his first 10 years after graduation doing VC, only to find that the academic SOP and actual investing were “completely unrelated.” Surveys, interviews, and statistics can decompose complex systems, but they rarely reach the variable that actually changes the decision when the trigger is pulled.
Lao Qian asks for the discipline’s core conclusion. 庄明浩 answers: “There is no core conclusion.” Theory can organize a problem; it cannot turn a complex system into a stable winning formula.
23. The push of a wave depends on where you stand in it
庄明浩’s first sense of being pushed forward came from mobile internet in 2010–2011. Smartphones in China surpassed 10 million units, the iPhone 3GS entered view, and the industry had not yet formed common rules. The information gap between young investors and their seniors was not as wide as it is today.
Blogs and Zhihu gave 庄明浩 a non-monetary form of positive feedback: industry analysis could genuinely affect founders and investors. He felt he was “rolling forward with the wave, perhaps even part of the wave.”
His second wave was game streaming. It was not everyone’s historical tide but the first time his interests, investment experience, and grassroots gaming ecosystem aligned tightly. “No one was more suited than you,” and he found his own small pond.
AI is so large that the individual becomes almost dust, while consensus is so strong that it leaves little room for peripheral participants to explore. Lao Qian calls himself “Sima Qian in the AI era,” able only to record. But the industry changes every two months, and the PPT must also be refreshed every two months; recording itself is already difficult enough.
24. Chinese investors’ silence weakens knowledge transfer outside the master-apprentice system
US VC partners, whether young or in their 60s and 70s, continue expressing views through articles, reports, podcasts, and conferences; institutions can even resemble media companies. Senior investors in China’s private and public markets, by contrast, are often “scared silent.”
庄明浩 attributes the difference to the opinion environment, the aura of righteousness, and the returns to speaking publicly. Some people demand that investors expose their taste; others label that expression in an inquisitorial way. The cost and benefit do not match, reinforcing silence.
The first time Chinese VC truly entered the mainstream was through Ma Yun, 熊晓鸽, 史玉柱, and others on Win in China. The public learned that venture capital meant “the money does not have to be repaid; it is not debt.” That positive narrative has since become more complicated.
25. Apprenticeship is deteriorating from close mentorship into solo combat
When 庄明浩 entered the industry, it was a strict master-apprentice system: a designated mentor took juniors to see comparable deals, while analysts wrote reports, joined meetings, and learned judgment. During his Matrix Partners China years, the model shifted to thematic groups in which partners, managers, and analysts collaborated around one direction.
Today, shorter project cycles, broader industry boundaries, and faster information updates have turned many teams into solo operators under loose structures. Having juniors write reports, record calls, or forward materials may not replace face-to-face transmission.
US VC investment teams look more like an inverted triangle: many partners, few junior investors, and a separate operations team. China has large numbers of analysts and managers supporting a small number of partners. The former makes it easier for a VP to follow a partner for years and eventually succeed them.
China’s first-generation VC partners have entered the handover stage, but trust, economics, and LP recognition are all obstacles. Some funds treat “completing generational succession” itself as PR, which only shows that the second generation does not rise naturally.
26. FOMO comes from rapidly exhausting a battlefield after it has been defined
庄明浩 describes private markets as a continuous battlefield: once a direction is summarized and its keywords become common, money from the core circle rushes in. Once top funds enter, institutions on the outside are forced to follow; the herd effect is unavoidable.
When top funds believe a direction may explode in three years, they can fully price in that future and buy as much allocation as possible at high prices, compressing a process that originally required multiple financing rounds. “No one else should participate,” after which capital moves on to define the next battlefield.
Private markets lack the instant matching of public markets. After buying, you wait a long time; the result may be zero or enormous. The winner’s curse is therefore not a short-term mark-to-market loss after overpaying, but discovering years later whether you prepaid for the entire future.
27. Fund inflation is upstream of company-valuation inflation
Lao Qian asks which is inflating faster, company financing or fund fundraising. 庄明浩 chooses fund size because fundraising follows macro liquidity while the supply of projects has not increased in parallel.
A team that once managed RMB200M and wrote checks of several million yuan may still target familiar stages and a similar number of companies after expanding to RMB800M–RMB1B. It can only increase check sizes, which in turn lifts valuations.
If competing funds all double in size, prices at the same stage rise as well. If you do not expand with fundraising, you are pushed down to the next table. Capital and talent become even more concentrated, and industry concentration rises sharply.
He cites US market data showing that the Top 10 funds take roughly 50% of fundraising and the Top 25 take roughly 85%. China has an additional liability-side problem: the sources of capital are “completely mismatched” with VC’s long cycles, tolerance for failure, and exit dynamics.
28. Effective information is inevitably copied; the moat often comes from a larger upstream force
Lao Qian uses podcast clips as an example: one program gained 300,000 subscribers in a month through the format, after which competitors copied it collectively. He moved from “not being ashamed to copy” to acknowledging that once effective information is validated, rapid imitation is simply how market competition works.
Game-streaming companies were once highly similar. In the end, the winner was determined not by who thought of the idea first but by which company Tencent and other upstream forces directed resources toward and acquired. A small pond is often prematurely matured by larger capital or traffic platforms.
A recognized rule inside VC is that when five or six companies receive funding for the same model at once, competition usually becomes extreme very quickly. Group-buying wars and community group-buying were both heated by execution, offline promotion, tactical optimization, and capital supply.
A beef-noodle project once reached a valuation of RMB100M. The prevailing logic was standardization, offline chains, new consumption, and private-domain operations. In hindsight it looks nearly insane, but people inside the market could not step back because every competitor was adding money.
29. Top funds cannot repackage non-participation as skill
Before leaving Matrix Partners China, 庄明浩 was responsible for the metaverse. He had watched gaming, social, 3D, and VR for years and could see that the underlying conditions had not changed enough. But once the keywords formed, companies with almost nothing could quickly reach valuations of several hundred million dollars.
He was unwilling to make a large bet when there was “nothing there,” but others would invest. Six months later, his year-end review showed no investment, and organizational pressure followed. For a top fund, saying no to a trend in which the entire market is participating may itself be a mistake.
Three years later, not investing may appear to have saved a lot of money, but that was not an answer available at the time. “A dog could say no there.” VC does not win by avoiding losses; it wins when a small number of shots hit.
30. Institutional taste about people emerges from the collision of multiple factors
A first meeting with a founder usually lasts only an hour, with roughly 40 minutes of founder presentation and 20 minutes of Q&A. Yet the investor must bet on the founder’s choices over many future years and across different market environments, while also accounting for industry, technology, and capital prerequisites.
If “continuous iteration, evolution, and vitality” cannot be measured, 庄明浩 calls it “bullshit.” Yet a fund must still summarize some basis for judgment, so colleagues, partners, GPs, post-investment, finance, and HR leaders meet the person separately and track whether each encounter adds or subtracts from the judgment.
Some institutions deliberately bring unrelated teams into a deal to increase sample diversity and avoid a shared group bias. Whether this truly improves the odds is unprovable, but everyone’s weighted judgment ultimately shapes the type of founder the fund prefers.
Serial founders and people who are not short of money often receive extra credit because some of their traits have already appeared across ventures. Their distinctiveness, however, remains a black box that cannot be reduced to a reliable scorecard.
31. Values can be neutralized; investors may still be unable to betray themselves
Lao Qian offers a sports analogy: if someone has terrible private morals but can score, should you still pass them the ball? 庄明浩 acknowledges that the market may prove him wrong, but in the first person he can only make choices he can live with.
Between a “decent good person” and someone “willing to do anything to win,” he clearly chooses the former because he cannot persuade himself to invest in the latter. Even if the latter can defeat an opponent through small tricks, that too is the investor’s choice and fate.
Early funds once had taboos against one-to-one replication and ranking manipulation. But the first entrant is not necessarily the ultimate winner; a copier may win through better localization. Whether taste correlates with worldly success has no universal answer.
Lao Qian insists on “passing the ball to the person who can score.” 庄明浩 does not deny the strategy: some investors can back almost anyone, and there are successful examples. It is simply not his way of operating.
32. Experience rules are scars, not general laws
庄明浩 keeps one hard rule: if a founder explicitly says, “If you invest in me, you absolutely cannot invest in competitors,” the fund does not invest. He reached that conclusion after three or four similar cases that were “all wrong, not one right.”
He also acknowledges that the sample is small and the rule may not be universally correct. The fund has simply decided to write its historical scar into the process and follow it collectively, rather than claim to have found a universal causal law.
Early funds held regular off-site discussions focused not on grand visions but on personality, human nature, preferences, and why a particular principle had been written into the decision process. Investors oscillate between the void and the details, adjusting continuously.
33. Diligence exposes risk while also participating in self-persuasion
Early-stage companies are inherently incomplete, so diligence will inevitably uncover many problems. When training analysts, 庄明浩 stressed that the deal lead’s job is to balance risk identification with momentum, not to seize every flaw simply to prove seriousness.
External lawyers and auditors have been paid to surface problems. The investor must understand their professional opinions while deciding which risks could change the transaction and which are normal for an early-stage company.
Extreme problems must of course be addressed, but diligence is not meant to reduce risk to zero. Requiring early-stage companies to have the completeness of mature businesses would itself eliminate many possibilities.
34. “You bet, you lose” is being redefined by serial-founder provisions
When 庄明浩 received his early education from dollar VC, the industry believed in “taking one big swing” and “you bet, you lose.” If more than 90% of projects might fail, normal failure should not become a permanent claim against the founder.
Agreements now often include serial-founder provisions: if the founder starts again after a failure, the original investor may retain a certain interest or share. This reflects the influence of renminbi capital and historical lessons, as well as the growing number of serial entrepreneurs.
A failed project need not mean a failed person. The same founder may be viewed as garbage by Fund A and a star by Fund B; whether the former invests again reflects only its own judgment and cannot shut off the founder’s next financing round.
Huang Zheng, Zhang Yiming, and Wang Xing had all tried other ventures before their major projects. The show keeps them as intuitive examples: failing on the first, or even the first several, attempts is not unusual in the entrepreneurial distribution.
35. Cleaning up after investments is the bulk of VC work
庄明浩 describes a deal he handled: of five companies in a hot sector, his fund invested in one, and the industry later shrank to two. The leader received a ticket onto a larger platform; if the second company did not get on board, the entire battlefield might disappear quickly.
The leader was willing to acquire the second company, but used its extremely high valuation for an all-stock deal. Even if the businesses were roughly 2:1 in scale, the exchange ratio could become 20:1. The second founder still accepted because a 5% ticket on the ship was more rational than remaining on the old battlefield.
As the company VP, 庄明浩 supported the founder, while partners and the boss above him might view the deal as a humiliating exit. Finance, legal, and IR then questioned it from colder functional perspectives. Managing these interests, emotions, and layers is “meeting soldiers with generals and water with earth.”
Later, the leader changed strategy and no longer pursued an IPO. The larger-company shares received by the smaller company were diluted down to a few ten-thousandths, with no decision rights and no clear exit. Which matters more—the sector, the horse, or the rider—still changes with stage and fund preference.
36. Without animal instinct, early-stage investing loses its ability to shoot
庄明浩 believes early-stage investing must retain the animal instinct to be bullish, go all in, and “take one big swing.” Curiosity may come from nature and enthusiasm; animal instinct is bloodier, more direct, and more tightly linked to results.
Some people’s utility curves decay slowly, so wealth and age do not stop them from looking at deals. 庄明浩 gives people considering the industry two screening criteria: the team should not be too large, and more importantly, “the boss should still be looking at deals.”
If the boss no longer looks at deals, the fund can easily become only organization, process, and hierarchy. Treating a VC institution like a startup, whether its founder remains close to the frontline is still an important signal.
37. Nobody really looks at the Gartner curve, but AI’s “inflection point” still demands a bet
Asked how the industry views the Gartner curve, 庄明浩 first gives the standard answer about different stages corresponding to different expectation functions, then tells the truth: “Nobody looks at that thing. Nobody cares.”
Lao Qian adds that a fund’s position determines which part of the curve it sees: late-stage funds look at the growth segment after the curve stabilizes. The tool does not replace the decision; it merely supplies a language for explaining an existing position.
庄明浩 refuses to choose between AI being in the introduction phase or the expansion phase because “one hell” lies between them. Lao Qian calls 2025 the year of the “inflection point,” emphasizing that an inflection point can go upward or downward.
The materials Lao Qian shows include sharp upward curves at the end of 2025 for Oracle’s expected cloud revenue over the next five years, OpenAI’s expected revenue, Nvidia’s data-center growth, and company valuations. 庄明浩 says that if this is the final 6–12 months of a bubble, 2026 could also bring an equally steep downward adjustment, while emphasizing, “I don’t know.”
38. Market judgment ultimately comes down to position size, tools, and the knowledge-action gap
庄明浩’s preference may persist for a while, but he believes a large correction is inevitable. His stress test is Nvidia falling roughly 40% from $180–$190; it had traded around $80 earlier this year, so a seemingly enormous decline might not break out of the year’s prior range.
Gold and equities hitting record highs at the same time invalidate the instinct that “one side must be wrong.” Lao Qian, looking at a Bitcoin cycle of three years of gains followed by one year of drawdown, leans toward 2026 potentially becoming a drawdown year.
Both acknowledge that anyone can find a historical chart that appears to support their view absolutely. Even knowing that the final phase of a bubble may be the most profitable does not mean one will dare to bet. “Knowing does not necessarily mean doing” is the hardest part of investing.
Lao Qian suggests reducing participation to a 2% position, analogous to an early-stage investment. But the odds on the common stock may not be attractive enough, leading him to ask, “Why not buy puts?” This is not an answer to the bubble; it is an attempt to translate the unknowable into odds one can bear.
39. People who think less may sometimes preserve the final needle more easily
庄明浩 has met excellent private-market investors who appear to “think too little,” or show only the final needle. They seize one heavily weighted judgment and fire, without exposing the full complexity.
Outsiders cannot tell whether that needle came from a watertight chain of reasoning or simply from deciding that something felt important. People who agonize repeatedly may have more information but are not necessarily closer to the correct action.
Following investors who express themselves publicly also has limits. Few Chinese investors are willing to speak openly in the first place; the more you watch, the more you realize that every statement has a position, purpose, and omission. It is difficult to treat anyone as someone to follow blindly.
Bezos can cite Amazon’s fall from more than $100 to $6 as evidence for his own framework, but strong experience does not mean a one-dimensional conclusion necessarily applies today. Investors can only choose which probability distribution to believe.
40. Two-month industry cycles coexist with five-year and 25-year narratives
Lao Qian recalls seeing the purple Podcast app while using an iPod touch in 2009, yet not opening it for more than a decade before eventually making podcasts his profession. Technology’s arrival, personal adoption, and commercial realization can be separated by a very long time.
The end of 2025 means the first quarter of this century is complete, and people instinctively push expectations toward 2030 and 2035. But the AI ecosystem turns over every two months, creating an unusually tense collision between long-term planning and ultra-short technology cycles.
庄明浩’s AI presentation moves from “is it a bubble?” to “what kind of bubble?”, “what do we do even if it is?”, and “how extreme would the break be?” The goal is not to choose an answer for the audience but to lay out the future probability distribution as fully as possible.
What he is willing to do is “unfold the whole pie,” showing the factors within reach. The final result may come from an unlisted black swan or a small variable suddenly turning from positive to negative.
41. Age makes people want to leave the acceleration zone; work forces them to keep pace
庄明浩 is approaching 40 and physically feels unable to keep up with AI’s extreme speed. But habits formed over more than a decade, along with writing and podcasting, require him to keep copying that speed.
Choices when young look more like 0 or 1. Today, family, work, opportunity cost, and the body must all enter the same table. “You can only adjust slowly in the middle”; there is no one-time solution to the tension.
During the pandemic he taught himself video editing and moved from participating in other people’s expression to producing his own podcast. Many decisions were not leaps of faith but the accumulation of multiple stimuli until they crossed the threshold of “I will not do this.”
He believes there may be only one or two opportunities in a lifetime to leap into a wave, and fewer in the future. The more people understand their own limits, the more variables they must balance.
42. Time moves at different speeds in Shanghai, AI, and the Northeast
Lao Qian experiences Shanghai and AI as high-speed bubbles, while returning to his hometown in the Northeast reveals streets and family routines that have barely changed in years. The same language, similar cars, and similar shopping malls can still coexist with two completely different subjective speeds of life.
庄明浩 says his wife experienced the same rupture when she took their children back to Shenyang. Excessive efficiency and excessive slowness can both produce a sense of emptiness; people at either end of the spectrum can experience crisis.
After short-video platforms make one content format explode, creators copy it collectively and users quickly become bored. The battlefield moves from definition to exhaustion with almost no middle ground. When Lao Qian saw podcast-clip traffic disappear, he was relieved: “Great. I didn’t want to make them either.”
The value of podcasts may not be the absence of emotion but sufficient length, with a beginning, development, and resolution. When short-form content reaches an extreme, some people develop a physical aversion and seek slower media. This is what 庄明浩 calls “when things reach an extreme, they reverse.”
43. VC remains a good job because it forces continuous input
Although its freedom is lower than outsiders imagine, VC gives individual judgment more weight than many industries. The people one meets, the excitement of the projects, and the connection to the pulse of the era still make the job feel sexy.
The work also requires input by nature: without continuously reading about technology, industries, and capital, one cannot judge deals. If someone is also suited to produce output, input, organization, and expression form a self-reinforcing growth loop.
庄明浩 always carries the label of “taking one big swing,” which he calls the industry’s dragon-slaying art. A profession built around odds does not disappear immediately when one leaves the frontline; age and family simply reduce its weight.
He stores materials on S Notes, covering AI, VC, gaming, social, and capital markets, then reprocesses them through PPT. The library can be shared, but what matters to him is the reworking, not mechanical backup.
44. Liquidity is both a macro cycle and a micro asset of long-term trust
Macro liquidity is shaped by interest rates, public markets, and risk appetite. Micro liquidity depends on the fund’s own historical performance. Even in the most pessimistic periods, institutions with genuinely sustained returns may still raise money.
The 10–12-year cycle weighs on liquidity but also allows trust, once established, to last more than 20 years. Long-term cooperation among university endowments, pension funds, and VC managers rests on a shared understanding of how this table is judged.
Even if US–China relations deteriorate, China’s top dollar funds may still raise dollar capital, though policy restrictions will apply. 庄明浩’s explanation is that as long as LPs believe technological innovation will continue producing extreme returns, they cannot abandon the market simply because of screening.
45. Fund strategy is a coordinate system; choose the wrong pond and you may survive only one vintage
A fund’s ability to survive may be defined by geographic, industry, temporal, and personal-network coordinates. These filters must match the evolution of the real economy; choose the wrong coordinates and the fund may be eliminated outright.
When entertainment was hottest in 2015–2016, investors in the sector could often raise a first fund on gaming-company exits and capital from listed companies. By the second fund, entertainment might no longer be a private-market battlefield. The network remained, but the opportunities disappeared.
ZhenFund’s overseas-returnee network is difficult to copy because Xu Xiaoping and Wang Qiang built genuine relationships and judgment through years at New Oriental. Other funds can imitate the surface but not the resources, trust, and sense of proportion.
Every advantage has a lifespan. A strategy that remains effective for years is not permanently stable. But if time could run backward, people who had already received positive feedback and assembled the necessary resources would probably still leap, because “fate pushed you there.”
46. Hard work is the entry ticket, but research effectiveness cannot be proven
When 庄明浩 was a frontline investor, he saw an average of two or three new deals a day—more than 200 working days a year—while managing existing investments, writing reports, attending weekly meetings, and handling post-investment work. This was close to the industry’s required routine.
The two mentors who trained him—吴峰, now at Northern Light Venture Capital and formerly at Shanda, and a former East China head at 经纬—both told him the same thing: “Hard work is only the foundation.” Without that intensity, talent, luck, and taste have no chance to operate; with it, returns are still not guaranteed.
Many Chinese VCs wanted to establish research institutes at the peak, but 庄明浩 says almost none of the independent research departments truly lasted. Public markets provide daily prices and returns as rapid feedback; private-market research may have no comparison point for years.
Worse, once a private-market battlefield ends, the previous round’s keywords, materials, and information may all become garbage. Analysts increasingly resemble assistants arranging meetings and taking minutes. Research and final returns are “of unknown relationship, but definitely not positively related.”
47. “Do what you can and leave the rest to fate” means building a daily routine, not pretending to control outcomes
庄明浩 believes in “do what you can and leave the rest to fate,” but only if the work is genuinely done—not if laziness and surrender are packaged as equanimity. For him, doing what he can begins with uninterrupted daily input and organization.
No matter how busy, late, or tired he is, he goes through every Substack subscription. He may not read each article deeply in the moment, but without organizing the information, much of it disappears from his mind two months later.
The habit cannot prove that it improves investment returns, but it forms the basis for people seeking his advice and companies listening to his views on business trade-offs. “Carrying these stones on the road is exhausting,” but he still cannot put down the materials he has already seen.
48. Education decisions recreate investing’s complex systems and irreversible bets
庄明浩’s older son is about to move from Shanghai’s five-year primary school system into middle school. Whether to switch out of the mainstream system implicates the world order, educational methods, the child’s personality, and his future path; if no adjustment is made in the first or second year of middle school, later options will narrow sharply.
The child is already 11 or 12, and his learning methods and attitudes toward his parents and future are taking shape, with many changes becoming irreversible. There is also a five-year-old daughter who will be influenced by her brother. Parents struggle to distinguish which experiences are mistakes and which outcomes came from “planting willows without meaning to.”
He took his son to an AI hackathon not as a teaching activity but as an exhibition and a way to cultivate taste: drip-feed the child whatever resources he understands, while admitting that this is also largely psychological comfort for the parents.
Every sentence and everyday decision in parenting accumulates, yet a random stimulus may still push the child past a threshold. As in investing, people can list A, B, and C in full but cannot ensure that the moment of pulling the trigger is not a snap judgment. All they can do is do what they can—and acknowledge fate.