The Optionality of Scarcity

In the boom years, a Chinese entrepreneur named Wang Ronghui raised venture money to build a chain of childcare centres. She signed the term nearly every founder of her cohort signed: a redemption clause, giving her investors the right to demand their capital back, with a premium, if the company failed to deliver an exit on schedule. Her investors told her the clause was a formality. They had never enforced one. As the Financial Times later reported, in 2017 and 2018 that was true — nobody was enforcing them.

Then her business stumbled, the IPO window that was supposed to provide the exit slammed shut, and the formality was enforced. She now owes her investors millions of dollars she does not have. Her name sits on China's national debtor registry. She cannot book the better hotels, cannot fly, in some cases cannot board a high-speed train, because the system flags her at the gate as a person who failed and could not pay for it. The equity she thought she had raised turned out, under stress, to be a personal loan secured against the rest of her life — and in a country with no general personal-bankruptcy discharge, there is no proceeding that ever clears it. The debt, and the blacklist, are for as long as they go unpaid, which on those sums is forever.

She is not an outlier. She is the leading edge of the largest and least-understood natural experiment in the history of startup formation. And the reason it is the most useful thing Indian deeptech could be looking at is that almost nobody — neither the people who think China is an unstoppable innovation juggernaut nor the people who think China is collapsing — is looking at the specific thing that happened to her, which is mechanical, nameable, and portable. What China did, without quite intending to, was repeal a piece of financial technology so old and so load-bearing that almost nobody names it anymore: limited liability. Not by statute — China's revised Company Law, in force since July 2024, leaves limited liability formally intact — but in practice, by stacking contract and enforcement on top of the company form until the cap no longer holds. This essay is about that de-facto repeal — what it has already done to China, why you have not seen it, and why it is the exact mistake India is one term sheet away from importing.

The China the headlines sell

There are two pictures of China in circulation, and they are at war with each other. In the first, China is winning. A Hangzhou lab ships DeepSeek and rattles the entire American AI industry. BYD passes Tesla. China builds more solar than the rest of the world combined, dominates batteries and drones and electric vehicles, and — as the companion to this essay documents — out-licenses roughly $136 billion of novel drugs in a single year while one of its antibodies beats the best-selling medicine on Earth in a head-to-head trial. In the second picture, China is sinking: a property crash that erased the largest store of household wealth in the country, youth unemployment bad enough that the government stopped publishing it, deflation, demographic decline, capital flight. Foreign direct investment fell from roughly $344 billion in 2021 to a thirty-year low, turning net-negative in the third quarter of 2023 and sliding to around $15–19 billion by 2024.

Both pictures are real, and both miss the thing that matters most, because both are arguments about output — about whether the machine is producing impressive things or faltering ones this year. The thing neither sees is what has happened to the machine's intake: the rate at which China starts new companies at all. And on that one number, the most important number in any innovation economy over a twenty-year horizon, the picture is not "winning" and it is not "slowing." It is a near-total stop.

The China the data shows

China minted 51,302 new venture-backed startups in 2018, at the height of its boom — well over five hundred new companies a day. By 2023 that figure had collapsed to 1,202. In 2024 it was tracking toward roughly 260 — for the entire year, in a nation of 1.4 billion people. That is a decline on the order of ninety-nine per cent in the rate at which the country forms new companies. "The whole industry has just died before our eyes," a Beijing executive told the FT. "The entrepreneurial spirit is dead."

The obvious objection is that China still registered more than ten million new companies in 2023 — so where is the collapse? The answer is in the denominator. Those millions are overwhelmingly small businesses, sole proprietorships, and self-employment vehicles. The figure that matters for frontier innovation is the rate at which venture-backed companies form — the ones that raise institutional capital to attempt something hard — and that is the number that fell ninety-nine per cent.

The capital that funds formation imploded in parallel. Dollar-denominated venture fundraising fell from a 2022 peak of $17.3 billion to under one billion. Yuan-denominated fundraising fell from roughly ¥125 billion in 2017 to a few billion. Total venture investment into Chinese startups dropped more than eighty per cent from its 2021 peak.

This is the picture under the picture. The DeepSeeks and BYDs are not evidence against it; they are its alibi. They are the harvest of a planting season that ran from roughly 2010 to 2021 — companies founded, funded, and grown when formation was still happening — and a spectacular harvest can run for years off a great planting season even after the planting has stopped. What the output cannot show you is that the field behind it is now nearly empty. The question this essay forces is the one the headlines structurally cannot ask: why did the most ambitious entrepreneurial culture on the planet stop making entrepreneurs?

The machinery of personal ruin

The answer is not that China ran out of talent, or ideas, or even money — state money, as we will see, is everywhere. The answer is that China made starting a company personally catastrophic to fail at, and rational people respond to that exactly the way you would expect: they stop.

The instrument was hiding in plain sight in nearly every term sheet. Redemption rights, and their cousin the valuation-adjustment mechanism that Chinese founders call 对赌 — literally "betting against each other" — were standard furniture. Shanghai law firm Lifeng Partners estimates more than eighty per cent of China's venture and private-equity deals carried them; in the US such clauses are comparatively rare. The name 对赌 is the tell the whole system tried to forget: this was never equity. It was a bet the founder placed against their own investor, and the collateral was the founder.

For a decade the bet was never called. Capital was abundant, exits were plentiful, and enforcing a redemption clause would have poisoned a founder's next round, so the clauses sat dormant and everyone agreed to pretend they were a formality. Then three things failed at once — the economy slowed, the IPO market froze, and the limited partners behind the funds began demanding their money back — and the dormant clause became the only asset a fund could realise. Investors enforced what they had promised they never would. Lifeng estimates a fifth of all investor exits in 2021 and 2022 already came from companies buying back their backers' shares, and that more than ten thousand VC- and PE-backed Chinese companies now face redemption claims. Founders who had built real businesses and merely failed to time a listing found themselves owing sums they could not pay. As one adviser put it to the FT, in the five words that are the whole repeal: it is not venture, it is debt.

And here is where China's machinery becomes something the West has no equivalent of, and the part the founder-ruin stories rarely connect. When the court rules and the founder cannot pay — and fewer than one in five who lose these buyback suits can, by Caixin's count — the founder is entered onto the Supreme People's Court's List of Dishonest Persons Subject to Enforcement: the 失信被执行人 registry, the laolai or "deadbeat" blacklist. This is not a metaphor for reputational damage. It is an operational, automated, nationwide system of exclusion. By 2018, around twenty-three million people had been blocked from buying plane or high-speed-train tickets under the wider social-credit system this blacklist feeds. Some 8.3 million sit on the debtor registry itself. In 2024 alone, roughly 2.46 million people were added. The blacklisted cannot fly, cannot take the fast trains, cannot stay in good hotels, cannot win government contracts; their names and ID numbers are published; in documented cases their children have been blocked from certain schools. And because China has no general personal-bankruptcy law — no Chapter-7-style discharge that lets an individual fail, clear the slate, and begin again — there is no exit from this state except to somehow pay. The founder is not benched. The founder is removed from economic society, indefinitely.

Stack those five and you have not a harsh culture but a designed machine, and its output is a single broadcast signal to every scientist and engineer still deciding whether to start something: build, and if you fail, we will not merely take the company. We will take you, your mobility, your children's options, and the rest of your economic life, and there will be no proceeding that ever gives them back. You do not need 1.4 billion people to do the expected-value math on that. Two hundred and sixty of them did it in 2024 and started anyway.

The hand on the switch

It is tempting to read this as cruelty, and it is not — which is what makes it so much worse, and so much more permanent. Nobody chose it. It fell out of a structure, and the structure runs all the way up to the state, which is the part almost no one outside China's venture world has connected.

Follow the money up the chain. The limited partners forcing redemption on the funds are, increasingly, the Chinese government itself. As foreign capital fled and private domestic funds dried up, the gap was filled by government guidance funds — state vehicles, central and local, that channel public money into priority sectors. They are now enormous: by the end of 2025 there were more than 2,100 of them with target capital exceeding eleven trillion yuan, and something like eighty per cent of Chinese venture capital now traces to state-linked sources. The risk capital that remains in China is overwhelmingly the state's.

Now add the fiscal fact that makes the machine bite. For two decades, Chinese local governments funded themselves substantially by selling land. The property crash of the early 2020s destroyed that revenue base, and local governments pivoted, hard, into equity finance — using guidance funds not only to direct industrial policy but as a hoped-for source of fiscal income. Which means the LPs at the top of the chain are not patient endowments that can write off a loss. They are cash-strapped local governments that staked public money they now need back, and that possess something no private LP possesses: sovereign enforcement power — the courts, the blacklist, the apparatus. The redemption wave looks, from the founder's seat, like greedy VCs. It is actually a fiscal emergency flowing downhill from broke municipalities through GPs onto founders, with the full machinery of the state behind the demand. That is why it does not get extended, renegotiated, or forgiven the way a private clause might. The creditor is the state, the state is short of money, and the state cannot be out-waited.

The state's other effect is quieter and just as corrosive. When eighty per cent of your risk capital is government capital, the capital stops behaving like risk capital. State funds answer to officials who must justify losses, so they crowd toward the safe and the directed — advanced manufacturing the centre has blessed, not the strange disruptive bet that defines venture — and fund managers who must, in the words of one analyst, dress their companies up as aligned with government priorities to get funded at all. The system has reorganised itself around not-failing. And an innovation system optimised around not-failing has, by construction, stopped doing the only thing that produces breakthroughs, which is making large numbers of independent bets most of which fail. China did not merely punish its failures. It replaced the capital that tolerates failure with capital that cannot afford to be seen tolerating it.

The larger game

The fiscal story and the crowding-out story could each be read as accident — the unintended fallout of a property crash and a liquidity freeze. Read alongside everything else the state has done, they look less like accident and more like the price of a choice. Because the same decade in which formation collapsed is the decade in which Beijing made unmistakably clear that it would rather have innovation it directs than innovation it merely hosts.

The signals were not subtle. In late 2020 the state pulled Ant Group's $35-billion IPO days before it priced — the largest flotation in history, cancelled from above. It opened investigations into Alibaba and into Didi, the latter just after Didi defied regulators to list in New York. After-school tutoring was forced into non-profit status overnight; gaming was throttled; and a succession of the country's most celebrated founders went quiet, stepped back, or vanished from public life for a stretch. The lesson every ambitious Chinese builder absorbed was that the state, not the founder, holds the wheel — and that a private fortune and a foreign listing are not so much assets as exposures.

Set against that, the venture collapse stops looking purely like a market failing and starts looking partly like a substitution. The guidance funds are not a stopgap for absent private capital; they are the preferred instrument — public money, centrally prioritised, locally deployed, aimed at semiconductors and AI and advanced manufacturing rather than at whatever a thousand founders might independently decide to build. A formation engine that throws up unpredictable, individualistic, foreign-capital-linked founders is, from the vantage of a state that prizes control, not an unalloyed good. The redemption wave handles the ones who fail. The larger game is about what happens to the ones who succeed.

Which is where Manus is the whole argument compressed into a single company. Beijing Butterfly Effect, founded by Xiao Hong in 2022, launched the Manus AI agent in March 2025; it went viral, topped a leading agent benchmark, and reached roughly $100 million in annualised revenue within eight months. In April 2025 the US firm Benchmark led a $75-million round at about a $500-million valuation — and Washington promptly opened a Treasury review under its new outbound-investment rules, with senators accusing Benchmark of arming China's AI race. So Manus did what a rational founder does when caught between two suspicious states: it ran. It relocated its headquarters to Singapore, laid off most of its roughly 120 mainland staff, moved its core team offshore, registered its parent in the Caymans, stored its data outside the mainland — the "Singapore-washing" playbook of Shein and TikTok — and recast itself as a Singaporean company with Chinese roots. By December 2025, Meta agreed to buy it for around $2 billion.

And then the other state moved. In April 2026, China's NDRC security-review office retroactively prohibited the Meta acquisition and ordered it unwound, explicitly mirroring Washington's own logic back at it. The message was the larger game stated plainly: a Chinese-origin company and its technology are national assets, and they do not get to simply leave. The founder who fails is deleted; the founder who succeeds is not free to exit. Both ends of the founder's option are clipped — the downside by the blacklist, the upside by the block.

This is the deepest version of the diagnosis, and the one neither the juggernaut narrative nor the collapse narrative reaches. China has not merely uncapped the downside of building. For its strategic winners, it has capped the upside too — you may not freely realise a great outcome by taking global capital or selling to a global acquirer, because the state retains a prior claim on the asset you became. An option clipped at both ends is not an option. It is a job, with extra steps and a blacklist for quitting. A system that has quietly converted entrepreneurship into a state-supervised post will get exactly as much entrepreneurship as you would expect from one — which, at 260 new companies a year, is precisely what the data already shows.

The technology nobody names

Step back from the specifics — the clause, the blacklist, the guidance fund — and the single mechanism underneath all of them comes into focus, and it is older than any of them.

Risk capital does not work without a floor under loss. This is not a moral claim; it is a structural one. The entire apparatus — venture funds, angel cheques, the willingness of a scientist to leave a salaried lab for a maybe — rests on one legal device that took centuries to invent and is now so ambient it is invisible: the principle that when a company fails, the people who backed it and built it lose what they put in, and not one rupee more.

Limited liability was genuinely radical when it arrived. Britain's Limited Liability Act of 1855 and the company law that followed were fought over precisely because critics believed a cap on downside would invite recklessness and fraud — that men would gamble wildly with other people's money if they could not be ruined for it. The critics were right about the mechanism and wrong about the verdict. A cap on downside does change behaviour. It makes people take risks they would otherwise rationally refuse. That is not a bug in the system; it is the entire function of the system. The cap is the product.

Everything in China's repeal stack — the redemption clause, the personal guarantee, the laolai blacklist, the missing bankruptcy discharge — does one thing: it reaches past the company's balance sheet into the founder's personal one, and converts an equity instrument, whose loss is bounded by definition, into a debt instrument, whose loss is bounded only by what can be seized and how long a life can be encumbered. Under calm conditions the conversion is dormant. Under stress it triggers, and the founder discovers that the thing they signed was never risk capital at all. It was a personal loan wearing an equity costume. China did not run out of entrepreneurs. It quietly switched off limited liability, and limited liability turns out to be the device the entire engine was running on.

Why it does not self-correct

A cyclical downturn ends. This will not, on its own, and the reasons are structural — which is the part that should reorder how seriously you take it.

The clauses are embedded: more than eighty per cent of a decade of deals carry redemption rights, a standing inventory of dormant defaults that detonate on any missed exit. There is no discharge: with no general personal-bankruptcy law, a ruined founder cannot clear the debt and re-enter, so the stock of removed founders only accumulates. The enforcer is sovereign and broke: the state-LP chain has both the need and the power to keep clawing back, and cannot be waited out the way a private fund can. And the signal is self-reinforcing: every blacklisted founder is a data point that lowers the expected value of starting a company for everyone still deciding, which suppresses formation, which thins the next cohort, which produces fewer of the survivors whose success would otherwise rehabilitate the signal. A formation engine is a confidence machine, and confidence, once a system demonstrates it will take your life for a failure, does not return on a schedule. China spent thirty years building the belief that it was safe to try. It unwound that belief in three, and beliefs of that kind are far cheaper to destroy than to rebuild.

This is the difference between a recession and a wound. The output — DeepSeek, BYD, the licensing harvest — will keep arriving for a while on the stored momentum of the old planting. But the intake has been structurally damaged in a way that compounds, and by the time the thin harvest of a stopped planting season becomes visible in the output, the cohorts that would have prevented it were never formed.

The mirror: India runs the same repeal, through a different instrument

Now turn the mirror, because the reflexive Indian read — China is cruel, we are not, this is their problem — is both too lazy to be useful and factually wrong in the way that matters.

On the surface India looks immune, even opposite. While Chinese listings froze, India's public markets boomed: 42 technology companies went public in 2025, up seventeen per cent on 2024, absorbed largely by domestic institutional and retail capital rather than fickle foreign money. The specific trigger of China's catastrophe — a slammed-shut IPO window with no domestic buyer of last resort — is simply not India's present condition. On that axis India is structurally better positioned than China was.

But deeptech is a different animal from the consumer and fintech listings driving that IPO count, and its numbers are sobering.

That cumulative $11.2 billion — everything Indian deeptech has ever raised — is roughly what a single late-stage American AI round now costs. India is not going to win deeptech by matching anyone's capital. That race is lost on arithmetic before it starts.

And India already runs an uncapped-downside regime; it just runs it through a different instrument. India does not mostly ruin its venture-backed founders through redemption clawbacks. It ruins them, when it ruins them, through the promoter personal guarantee. India's is a promoter-led business model whose reflex is that founders personally guarantee their company's debts. The country has a modern bankruptcy law — the Insolvency and Bankruptcy Code of 2016 — but it is built to protect creditors and distrust founders, not to give them a clean exit. Under Section 29A(h), once any creditor invokes a founder's personal guarantee, that founder is barred from submitting a plan to rescue their own company. And since a 2019 notification upheld by the Supreme Court in 2021, creditors can push the guarantor into personal insolvency, putting house and savings on the table. The scholarship on Section 29A reads it plainly as a structural distrust of founders whose companies failed ever regaining control.

Different instrument, same repeal, same deleted founder. And the forward risk is the convergence point. India's deeptech is capital-scarce and exit-thin — precisely the pressure that, in China, pushed equity investors toward debt-like clawbacks in the first place. If Indian deeptech tries to compensate for thin returns by loading term sheets with redemption rights and guarantees that survive into personal balance sheets, it will manufacture China's outcome on a smaller base, just as its first real generation of deeptech founders is forming. The trap is not inherited. It is available, and scarcity is the thing that baits it.

The optionality of scarcity

Now the economics, which is where scarcity stops being a handicap and becomes the argument.

A founder operating under genuine limited liability holds, in financial terms, a call option on their own work: loss bounded to what they put in, upside unbounded. This is the same instrument "The Venture Math of Biotech" places at the centre of the entire field — value made at a few uncapped tails and destroyed cheaply everywhere else — and the correct way to hold a position shaped like that is to want variance, to take the bold, strange, high-tail bet rather than the safe incremental one, because the asymmetry only pays if you swing. An ecosystem full of such options is, in Taleb's sense, antifragile: it gains from the volatility of many bold attempts, because each loss is clipped at the cheque size while each win runs uncapped. Uncap the downside and you flatten the option into a personal liability with a lottery ticket attached — at which point the rational founder stops swinging, and the system loses the exact tail events it exists to produce. China's formation collapse is what that flattening looks like at national scale.

Here is the part that should change how an Indian deeptech investor thinks about being poor. The convexity of a bet is set by its structure — bounded loss, open upside — not by the size of the cheque. You do not need more capital to make a payoff convex. You need only to guarantee the floor. This is the same point "The Studio Math of Biotech" reaches from the other side: the edge is owning the shape of the bet, not deploying the most money. Which means the single highest-leverage intervention available to Indian deeptech is also one of the cheapest things in the entire system — a legal and contractual commitment that failure stays capped at the capital and never reaches the founder's life.

When capital is abundant, you can be sloppy about this, because volume manufactures outliers on its own: spray enough bets and some land regardless of how you shaped them. That was China's old strategy — the fish strategy from "The Harvest Illusion," enormous numbers of offspring, almost nothing invested in each, win on volume — and it worked until the day the downside got switched on. When capital is scarce you cannot buy outliers with volume. The only way to produce them on thin capital is to maximise the convexity of every bet: many small swings, each with its downside clipped and its upside open — the mammal strategy, few offspring deeply tended. Scarcity does not merely permit the capped-failure strategy. It mandates it. The poorer the ecosystem, the more the cheap structural edge is the only edge there is.

That is the inversion in the title. Scarcity looks like the constraint. Under a genuine floor it becomes the forcing function that pushes you toward the one strategy — bounded loss, maximal optionality, the micro-bet repeated — that a richer ecosystem is wealthy enough to ignore and a crueler one actively destroys.

And note what China's larger game leaves on the table for India to pick up. The floor is only half the instrument. China clipped the founder's option at both ends — capped the downside with the blacklist, capped the upside with the block on exit — and an option clipped at both ends is the thing nobody rational will hold. India can offer the opposite, and it is the rarest thing in the world to be able to offer for free: an un-clipped option on both ends. Safe to fail — your life stays off the table. And free to win — take global capital, list where you like, sell to whoever will buy, keep what you built, with no sovereign holding a prior claim on the asset you became. China has just spent a decade demonstrating that it can no longer offer either end. India does not need a single additional dollar to offer both. It needs only to decide that it will.

What capping actually looks like

The prescription is concrete, and most of it is contractual rather than legislative, which means it does not wait on anyone.

Keep risk equity actually equity. The most important clause in an Indian deeptech term sheet is the one that should not be there: no redemption right, no buyback, no guarantee that survives the company's death into the founder's balance sheet. Risk capital that converts to personal debt under stress is not risk capital; it is a loan mispriced as equity, and it carries China's outcome inside it as a dormant default. A fund that strips these terms out is not being generous. It is refusing to write the off-switch into the floor.

Small cheques, many bets — convexity as a portfolio. A fund writing $200–400K cheques across a wide field has, by construction, bounded its loss per bet to something it can absorb dozens of times while leaving each upside uncapped. Boundless's range is, I would argue, accidentally exactly the right shape, whatever the reasoning behind it. Capital scarcity, which looked like the binding constraint, becomes the discipline that forces the convex portfolio: you cannot afford to over-concentrate, so you are pushed into the optionality-maximising structure whether you reasoned your way to it or not.

Build the soft landing as infrastructure, not sentiment. The point of a floor is not kindness; it is that the structure, not the founder, absorbs the loss, so the founder walks out solvent, reputation intact, free to do the one thing that makes the whole system compound: start again. India's missing pieces are identifiable. One is the second cheque for the good founder of a failed company — the explicit willingness to back the person across the grave of the first attempt, which barely exists in a market this young. The other needs actual reform: a failure path that does not, via Section 29A and the guarantee regime, legally delete the founder the moment things go wrong. You cannot build a serial-founder ecosystem on a code designed to bar the promoters of failed companies from ever holding the wheel again.

The asset volume cannot manufacture

All of this converges on a single asset, and it is worth being precise about why it is the one that matters.

The serial founder — the person on their third company, who has been expensively wrong twice and survived to apply what they learned — is the one input that capital cannot buy and volume cannot produce. China could spray fifty thousand first-timers a year and still not synthesise the pattern-recognition that lives in one founder who has failed, in detail, and come back. That judgment is grown only in survivors, and only in a system that lets its failures live. It is the human form of what "The Harvest Illusion" called the seed bank — variation and judgment held in reserve, worthless to harvest and decisive to keep — and, like a seed bank, it is destroyed the instant a system treats this year's yield as the only thing worth protecting. China's blacklist is, in this exact sense, a seed-bank incinerator: it takes precisely the failed-but-wiser founders who are the most valuable input to the next cycle and removes them from the game permanently.

India's deeptech is young enough that it has barely begun to accumulate these people. With on the order of 354 Series A+ companies in the entire sector, the country is roughly one generation of failures away from either having a stock of hardened, twice-burned, still-building founders — or not, depending entirely on whether that first generation of failures is preserved or deleted. That fork is being decided right now, in term sheets and in the enforcement posture around the guarantee regime, mostly by people who do not know they are deciding it. In an ecosystem this capital-poor, the compounded judgment of surviving founders is not a nice-to-have. It is the moat — the one edge that deepens precisely because capital stays scarce, since the thing it is made of is preserved for almost nothing and a richer, crueler system actively burns it.

The cheapest edge

India is not behind in deeptech because it is poor. The poverty is real, and it is the wrong diagnosis. The capital gap is unwinnable and also, on this argument, beside the point, because the value was never in the volume of capital — it was in the shape of the payoff, and the shape is set by structure, not by dollars.

China has just demonstrated, at the scale of a national economy, what happens when you let the downside become uncapped: not cruelty for its own sake, but a formation engine that fell ninety-nine per cent because limited liability had a hidden off-switch and a fiscal emergency flipped it, with a sovereign blacklist to make the flip permanent — and, for its few winners, a state that will not even let a great outcome leave the country. That is the China you have not seen — not the juggernaut and not the collapse, but the quiet, structural, self-inflicted clipping of the founder's option at both ends, downside and up. India has the same off-switch wired into a different instrument, the promoter guarantee, and the same scarcity-driven temptation to wire it into its venture terms as well, just as its first real cohort of deeptech founders takes shape.

The intervention that prevents it is nearly free. Cap the failure — keep the equity equity, keep the loss bounded to the capital, keep the founder's life off the table — and the same scarce dollars buy a convex payoff instead of a flat one, the same cautious scientists become rational risk-takers, and the same dead companies leave behind living founders who compound. The most expensive edge in deeptech is capital, and India cannot afford it. The cheapest edge is the one China just proved it could no longer afford to keep: a floor under failure, and founders left intact to use what it taught them.

That edge is not bought. It is chosen — one term sheet and one statute at a time. It is the rare advantage that gets more valuable the less money you have, which makes it, for an ecosystem as scarce and as early as India's, very close to the whole game.

The deep-companion to "The Harvest Illusion," and the capital-structure layer beneath "The Venture Math of Biotech" (the founder's call option; value at the uncapped tail) and "The Studio Math of Biotech" (convexity owned by structure, not bought with scale). China's formation and funding collapse — 51,302 new companies (2018) → 1,202 (2023) → ~260 (2024); dollar VC fundraising $17.3B (2022) → <$1B; FDI ~$344B (2021) → a 30-year low, net-negative in Q3 2023 and ~$15–19B by 2024 — is from the Financial Times via Fortune, Axios, TechCrunch and Slashdot, on IT Juzi and Preqin data (FDI via World Bank BoP and SAFE). The redemption mechanism (Wang Ronghui; "it is not venture, it is debt") is from the FT's January 2025 reporting; its scale (>80% of China VC/PE deals carrying redemption provisions, a fifth of 2021–22 exits via buyback, >10,000 companies exposed) from Shanghai law firm Lifeng Partners, with Caixin's finding that fewer than one in five who lose buyback suits can pay. The laolai blacklist (失信被执行人) figures — ~23M barred from flights/trains, ~8.3M defaulters, ~2.46M added in 2024 — are from the Supreme People's Court via ChinaFile, LegalClarity, IBTimes and others; the absence of a general personal-bankruptcy discharge is widely reported. The state-capital chain — government guidance funds (>2,100 by end-2025, target >¥11 trillion; ~80% of VC now state-linked) and the land-revenue-to-equity-finance pivot after the property crash — is from the China Quarterly (Cambridge), the BU Global Development Policy Center, KrAsia, CNBC and Rhodium Group. The larger-game material — the 2020–21 crackdowns (the cancelled Ant IPO, the Alibaba and Didi probes) and the Manus arc (Xiao Hong / Butterfly Effect; the Benchmark round and US Treasury review; the Singapore relocation and mainland layoffs; Meta's ~$2B acquisition; and the NDRC's April-2026 retroactive block and unwind order) — is from Foreign Policy, APF Canada, The Online Citizen, TMTPOST and Bloomberg/Semafor reporting. India deeptech figures — ~7,500 companies, ~1,720 funded, $11.2B raised across the sector's history, 354 Series A+, 4 unicorns, 99 acquisitions, 37 IPOs, ~$1B/year; 42 Indian tech IPOs in 2025 — are from Tracxn (via TechCrunch). The legal frame is the UK Limited Liability Act 1855 and India's IBC 2016, Section 29A(h), and the 2019 personal-guarantor notification upheld by the Supreme Court in 2021. All figures are dated and directional. The thesis is one line: limited liability is the technology the whole edifice runs on; China has shown, at the scale of a nation, what its quiet repeal costs; and the cheapest, highest-leverage thing a scarce ecosystem can do is refuse to flip the switch.