The PUNZI Papers Essay Nine · The AI Bubble and Circular Finance Ponzi

This essay is part of The PUNZI Papers, a series of essays explaining the reason for our project, PUNZI WARS, and the potential we think it points to.

Visit us: https://punzi.xyz Follow us on X: @PUNZIwars

Advisory: PUNZI WARS is a game and an on-chain research experiment. It is not an investment. Participate at your own risk.

The largest companies in history are valued on stories rather than on any profit, floated as an exit for the insiders who got in early, and lifted by an artificial-intelligence build-out whose revenue, on close inspection, is partly the same money travelling round a circle and counted as growth at every lap. It is the apex of the everything bubble, an entire industry built like a Ponzi, and it sits on no floor at all.

A ‘Sentient Sun’ at 1.75 Trillion Dollars

On 12 June 2026 the largest initial public offering in the history of capitalism opened for trading on the Nasdaq, under the ticker SPCX. SpaceX, having four months earlier absorbed Elon Musk’s artificial-intelligence company xAI in an all-stock merger, pricing its shares at one hundred and thirty-five dollars. The offering raised on the order of seventy-five billion dollars and carried an implied valuation near one and three-quarter trillion, roughly sixty-one per cent above the level the company had set in its own December tender offer six months before. It valued the merged entity at approximately ninety-four times its prior-year revenue, a multiple with no precedent among the most valuable companies on earth.

Set that against the financials the offering itself disclosed. Consolidated 2025 revenue was around eighteen and a half billion dollars, against a net loss of nearly five billion. The artificial-intelligence half of the business, the half supplying most of the narrative, produced about three billion in revenue and lost more than six billion at the operating line, on capital expenditure of roughly thirteen billion. So a company was valued at one and three-quarter trillion dollars while losing money at every level, and the segment carrying the story was spending three times as fast as it took in.

Set it, too, against the stated purpose. The animating mission of the merged company, in the words of the memo that announced the combination, is ‘scaling to make a sentient sun to understand the Universe and extend the light of consciousness to the stars’. That was the declared object of the most valuable new public company ever floated, and a generation of retirement savings was about to be pointed at it.

One does not have to take a satirist’s word for any of this, because the most credentialed valuers in the market said it in plain sight. Aswath Damodaran, the New York University professor known as the dean of valuation, put the company’s worth nearer one and a third trillion dollars, a quarter below the asking price, and that was after stretching every assumption to its limit. He called the twenty-eight-and-a-half-trillion-dollar ‘addressable market’ the prospectus claimed for itself a figure that ‘borders on fantasy’, and wondered aloud what an entirely Grok-written prospectus would have looked like, and whether anyone would have noticed the difference. Morningstar put fair value at around half the asking price and called the artificial-intelligence unit a material threat of value destruction. None of it mattered. The shares priced at a hundred and thirty-five dollars, opened at a hundred and fifty, and closed their first day near a hundred and sixty-one. Elon Musk became, on paper, the world’s first trillionaire. The experts said the number was a story. The market bought the story at whatever number.

There is one detail the coverage mostly treated as a technicality, and it is the one that should arrest the reader. Because SPCX is so large, its admission to the major equity indices is effectively automatic, and once it is in them every index fund and default retirement allocation that tracks them must buy it, not because a single buyer has judged the company to be worth ninety-four times revenue but because the rules require the purchase. The largest, most speculative, least-earning megacap flotation in history arrives pre-equipped with a wall of buyers contractually forbidden from asking what it is worth. That is not a market pricing a company. It is a company pricing itself into the savings of people who never chose it, by a machine built to buy whatever is large enough to be bought.

A Price With No Floor Under It

The previous essay in this series made the general case: the stock market itself is Ponzi-like, a structure in which earlier holders are paid by later ones and the price floats on the arrival of the next buyer rather than on profits distributed. This essay is about the place where that pattern reaches its most extreme and least disguised form. SPCX is a specific example, with the artificial-intelligence build-out being a broader one.

Take the three Ponzi-like features and push them to the edge. Earlier participants are paid from later ones: the venture investor is cashed out by the public-market buyer at the flotation, that buyer is lifted by the index fund, and the index fund is fed by the worker’s automatic contribution, each gain realised by selling to the next entrant at a price set by the next entrant’s arrival rather than by anything the company earned. Continued operation depends on continued inflow: both the premium valuation and the loss-making growth company require capital to keep arriving on schedule, because there is no stream of profit underneath large enough to hold them up alone. And the structure produces far less than it references: in SPCX we witness a share priced at ninety-four times revenue, valued on annual recurring revenue it has not yet recurred, whose actual ‘revenue’ is largely its own suppliers’ capital flowing back to it in a circle.

None of this is officially a fraud. It is the disclosed, legal – and yet blindly celebrated – operating mode of the most valuable corner of the contemporary market, which is exactly what makes it the most dangerous instance in the series. Every earlier structure, when pressed, pointed downward to the same bedrock: the real economy of companies actually earning money. This is the corner where that bedrock is thinnest, where price has come furthest loose from production, and where there is no firmer floor beneath to appeal to. It is the everything bubble in the precise sense: not a bubble in one sector the rest of the economy can absorb, but the over-pricing of the asset every other structure in this series was quietly assuming would hold.

Revenue That Goes in a Circle

The defining financial structure of the present moment is the circular financing of the artificial-intelligence build-out, the productive-nullity feature of a Ponzi-like structure made almost literal. The shape is thus: Nvidia, the dominant maker of AI chips, announces a plan to invest as much as a hundred billion dollars in OpenAI, the laboratory whose models define the boom (a non-binding letter of intent it has since stepped back from, which says something about how firm the figures in this circle are). OpenAI, in turn, committed to buy computing capacity on a scale no customer has ever contracted for: reported across 2025 at roughly three hundred billion dollars with Oracle, around ninety billion with Nvidia’s rival AMD, which sweetened the deal with stock warrants, and tens of billions more with others. That capacity is built, overwhelmingly, out of Nvidia’s own processors. One cloud provider constructing it, CoreWeave, counts Nvidia as both a minority investor and, through a capacity backstop, a customer of last resort, while its own contracts to serve OpenAI tie its fortunes straight back to the chip maker. Money leaves Nvidia as an investment, travels around a circle of suppliers and customers who are also one another’s investors and customers, and returns to Nvidia as revenue, revenue that lifts its share price, which funds the next investment, which starts the circle again.

Each individual deal is disclosed, lawful, and signed by serious people. What the circle does in aggregate is let the same dollars be counted as revenue at several stops along the loop, so that much of what is reported as the explosive growth of the AI economy is, on close inspection, capital changing hands among a small group of mutually invested companies and recognised as sales at each transfer. Analysts have begun, carefully, to describe it as exactly that: one company’s spending is another’s revenue, all of it resting on the conviction that end demand will eventually arrive to justify the capacity being built for it. The combined free cash flow of the largest builders is forecast to fall sharply under the weight of the spending. The end demand (the actual paying use of all this computation, at a price that covers its cost) remains a projection.

This is the Ponzi-like condition the sovereign-debt essay named, transposed into the equity market. The structure rolls forward only so long as the inflow continues, and the inflow continues only so long as everyone believes the demand is coming. It is not that these companies make nothing; they make extraordinary things. It is that the revenue growth their valuations rest on is, to a degree no one can cleanly measure from outside, the sound of the same money going around a circle, where each is lap booked as proof that the next is justified. And these valuations are not a corner of the market. They are the market: a handful of AI-exposed megacaps now make up so large a share of the major indices that the captive retirement bid (those buying the index) is buying mostly them.

The mechanism has an older name and a documented history: vendor financing, which built the last technology bubble as faithfully as it is building this one. A case-in-point is Cisco Systems, whose routers and switches the internet was then being assembled from. During its 2001 financial year, Cisco carried close to two billion dollars in loan commitments to its own customers, extended so they could buy Cisco’s equipment. The sale was booked as revenue, the loan sat on the balance sheet as an asset, and for as long as the customers kept ordering and repaying it all looked like growth. When the demand it had manufactured evaporated, Cisco wrote off some two and a quarter billion dollars of inventory in a single quarter, the largest such charge any company had recorded to that point, and the loans to buyers who no longer had businesses followed the equipment into the loss column. Cisco was never charged with anything and never restated a figure: the apparatus was legal, disclosed, and exactly as circular as it appears. A company that lends its customers the money to buy its product has not discovered demand. It has manufactured the appearance of it, and booked the appearance as a sale.

The contemporary circle carries a charge the telecom version did not. The chips at the centre of it are not a durable asset: a leading-edge processor installed today has a working life of a few years before newer silicon and ordinary wear make it uneconomic to run. The hundreds of billions spent to build the capacity are therefore not a one-time outlay but a meter running down, and the revenue meant to justify the spending must arrive not eventually but before the hardware it was spent on has depreciated to nothing. The circle is not merely booking the same money more than once; it is doing so on top of equipment that is quickly expiring underneath it.

The Exit Dressed as an Arrival

The previous essay named two of the ways a modern valuation is manufactured rather than earned: the company buying its own shares, and the business valued on a projected future of revenue it has not yet collected. This essay’s machinery adds two more, and they are the ones that run hottest at the apex.

The first is the exit dressed as an arrival. An IPO is presented to the public as a company’s debut, the moment the world is invited to share in its future. It is at least as often the moment the earliest holders arrange their departure: the venture funds, the founders, the pre-IPO insiders selling into the enthusiasm a roadshow is designed to manufacture. The price at which they sell is frequently a ‘valuation’ set in private rounds by other investors marking their own holdings, a number with no transacting market beneath it. What we have is mark-to-model dressed as mark-to-market. SPCX is the apex specimen, a private valuation lifted sixty-one per cent in six months and handed to the public at ninety-four times revenue, an exit for the early holders dressed as the public’s chance to buy the stars.

The machinery of the flotation serves that exit at every step. The shares are priced the evening before by the underwriting banks and allocated, overwhelmingly, to institutions and favoured clients before the public can buy a single one; they open to general trading the next morning, where they frequently ‘pop’. The pop is presented as the market’s roar of approval. Across 1999 and 2000, the years of the last great flotation mania, the money left on the table this way (the first-day gain multiplied by the shares sold) came to roughly sixty-three billion dollars, an average of some seventy-nine million an offering, foregone by the companies floating and captured by the allocated buyers. A lock-up then holds the earliest insiders in their seats for a fixed period, commonly around a hundred and eighty days, after which they may sell; the price routinely sags as the date nears, in plain anticipation of the supply about to be released. The public is invited to the celebration of an arrival and is, more often, the counterparty to a departure.

And the valuation the insiders exit at is frequently not a market price but a private mark: a number one set of investors wrote down about their own holding and a later set agreed to for reasons of their own. WeWork is the specimen the decade will be remembered by. A company that rented out desks, restyled as a technology platform, was carried in early 2019 at a private valuation of forty-seven billion dollars, a figure its largest backer had essentially authored; the offering document, when it appeared, disclosed a measure of profit the company had invented for the purpose (‘Community Adjusted EBITDA’) that added back the ordinary costs of running the business until a loss read as a surplus. Given its first clear look at the accounts, the public market declined the story. The flotation was withdrawn, and within weeks the same company was rescued at a valuation below eight billion dollars, a fall of more than eighty per cent from a mark that had never been more than an agreement among interested parties to call it so. None of it was illegal: the forty-seven billion was disclosed, the invented metric was disclosed, the collapse was seen.

A private valuation is an opinion in the costume of a fact, and the costume comes off the instant a real market is allowed to vote. That vote, moreover, is increasingly held later and later, or not at all. Companies now stay private far longer than they once did, financed through round after private round at valuations set by the same narrow circle of funds that hold the earlier rounds and have every reason to mark them upward, with a parallel private-credit market lending to them away from the disclosure a listing would require. By the time such a company reaches the public at all, the largest re-pricings of a story into a number have already happened in a market that reports to no one. The flotation that once began the public’s information increasingly ends it: the last mark in a long private chain of marks, sold to the buyer at the top of a staircase they were never permitted to watch being built.

The second machine is the newest and most literal, examined above: to manufacture not merely the price but the revenue, through the circular financing of the AI build-out, in which the same dollars are booked as sales at each stop around a loop of mutually invested companies. The two run together with the two from the previous essay on the stock market, so that the modern artificial-intelligence champion buys back its stock, is valued on a projected future, was floated as an exit, and books circular revenue, all at once.

Valued on a Promise

There is a quieter machine than the flotation, and it runs continuously rather than once. A large part of the modern market is valued not on what its companies earn but on a projection of what a customer today will pay across the years to come. The subscription business is its purest form. A software firm spends heavily now to acquire a customer and books that cost against an assumed lifetime of payments the customer is projected to make, and to expand, at retention rates the firm itself selects. The present valuation then rests entirely on assumptions about the future. An entire vocabulary grew up to state these assumptions with the confidence of measurements (annual recurring revenue, net revenue retention, the cost of acquiring a customer, the payback period, the magic number, the Rule of 40), and the vocabulary became its own camouflage, fluent to those who trade in it and opaque to everyone whose savings are pointed at the result. The revenue is called ‘recurring’ before it has recurred. The lifetime is booked before it has been lived.

For a while the market paid almost anything for the projection. By late 2021 the public software companies were valued at a median of roughly twenty times their forward revenue, a price defensible only by assumptions about customer behaviour that had never been tested through a downturn. When the interest-rate environment turned in 2022 the same companies were repriced at around five times revenue inside eighteen months, a fourfold compression that fell not on the businesses (which mostly went on operating) but on the assumptions the price had been built from. Three-quarters of the valuation turned out to have been a mood.

The pattern is not confined to software, and it is at its most naked when the story arrives ahead of any revenue at all. In 2013 and 2014 the shares of the leading makers of 3D printers rose several-fold on a conviction, ratified by the technology forecasters of the day, that a printer was about to appear in every home and turn out the household’s own goods on demand. At the peak the sector was priced for growth of forty to fifty per cent a year that never came; the printers did not arrive, the revenue disappointed, and the shares surrendered seventy to eighty-five per cent of their value over the two years that followed. When shareholders later took one of the companies to court over its promotional claims, the judges dismissed the case, holding the statements ‘so vague and such obvious hyperbole that no reasonable investor would rely upon them’. The market is permitted to price a thing on a story, the story is permitted to be hot air, and when the two are set into a share price it is not a scandal but an ordinary week. The greater-fool machine needs no one to lie. It needs only the next believer to arrive before the last one asks to be paid out.

The Squeeze

The previous essay described where the inflow comes from: the captive, price-insensitive bid of the index funds, fed every payday by the automatic retirement contributions of people who never look. At the apex that steady tailwind turns into something sharper. The bigger a company’s price grows, the more of it the captive bid is compelled to buy, and at the very top the compulsion becomes a squeeze.

SPCX is the demonstration, and it was not lost on the market’s sharpest observers. Bloomberg’s Matt Levine, in a column titled ‘The SpaceX Squeeze’, described the flotation plainly as a squeeze on the index funds: a company large enough to enter the indices must be bought by every fund that tracks them, at whatever price the rules admit, regardless of what it is worth, the passive vehicles forced to chase the very thing the dean of valuation had just called a story. A company can be carried to the apex of the index on a story, and the moment it is large enough, the index bid is compelled to fund it, channelling the automatic savings of a generation into the most speculative valuation in history without a single saver being asked.

The structure the pension essay described as a captive audience and the structure the previous essay described as a market that prices itself are the same structure seen from two ends. The worker is the inflow. The worker is also the holder who will be standing there when the inflow stops, because the worker is the last buyer in the chain, and there is no one behind them to sell to.

The compulsion is written into the rules of the indices. When a company grows large enough and clears the stated bars (a market value now above twenty-two billion dollars, positive earnings across its last four quarters, a sufficient share of its stock in public hands), the index committee admits it, and on the day of admission every fund built to track that index must buy it, at whatever price the market is showing, in proportion to its weight. The money bound by that promise is the part that should give pause: something on the order of thirteen trillion dollars sits in vehicles that must, by mandate, hold the index as it is drawn. When Tesla was admitted to the main American index in December 2020, the biggest company ever added at the time, the funds obliged to hold it had to buy close to two hundred and twenty billion dollars of its shares to bring themselves into line, an ocean of buying set off by a committee’s decision and not by any buyer forming a view on the price. The bid is not persuaded. It is compelled, and compelled to buy the most of whatever the story has already priced the highest.

The Three Axes

The structure carries the three Ponzi-like features. The question this series then puts to any such structure is which of them fail, and how badly, along three axes: efficiency, transparency, and longevity. On all three, the case at hand is the worst the series has examined.

Efficiency

The axis asks how much of the capital moving through the structure is taken by intermediaries before it reaches the participants the structure exists to serve. The market’s legitimate function is real, and in the index fund it is genuinely efficient. The failure is the now-familiar bundling: the same market that allocates capital productively also runs an enormous apparatus of extraction that rides on the productive part. The underwriting fees on the largest IPO in history. The spreads on the debt that funds buybacks. The executive pay indexed to per-share metrics a buyback lifts regardless of operations. The sponsor economics of a SPAC that pays whether or not the company survives. The management and performance fees on the private funds marking their own holdings up to the valuations the public is then sold. And one extraction is structural rather than skimmed: capital that buybacks return to shareholders is capital not reinvested in productive capacity, so the more the design succeeds, the thinner the real base beneath it grows.

Transparency

The axis asks whether participants can understand what they are inside, and it requires not mere legal disclosure but the second-tier kind a non-specialist could actually parse and act on. Everything here is disclosed, and almost none of it is parseable, which is the functional opacity this series has tracked through every institution. The share of an earnings-per-share increase that came from share-count reduction rather than from the business is reconstructible from the filings, and is never in the headline. A private ‘valuation’ is presented as a fact when it is an opinion the holders wrote about their own asset. The AI revenue circle is visible only to whoever can map the entire loop of mutual deals, which is no outside investor and arguably no inside one. And the captive bid is the purest case of all: the worker is told they own a diversified slice of the economy, which is true, and is not told that their automatic contribution is a price-insensitive buyer of last resort funding whatever the market has most inflated, which is also true. The disclosure is complete. What the disclosure is for is to be filed, not read.

Longevity

The axis asks whether the structure is built to survive the failure of its growth assumption. The growth assumption here is that the inflow continues: that capital keeps arriving, that the captive bid keeps buying, that the next funding round, the next index inclusion, the next enthusiast keeps appearing to lift the last holder out. It is the most tested assumption in the series and the one with the worst recent record: the 2022 software-multiple collapse showed that three-quarters of a valuation could be capital-market mood, and it repriced in eighteen months. The AI build-out is that same assumption inflated to a scale the others never reached, resting on an end demand that has not arrived and funded by a circle of mutual investment that books the waiting as growth. A structure whose entire stability depends on the inflow never pausing, in a market that can reprice three-quarters of a sector’s value in eighteen months, is not built to survive the pause. It is built to postpone it, and to hand the cost to whoever is holding when it ends.

The Story Is the Product

The three axes are three readings of one condition: a functional opacity that hides how little production stands beneath the price, laced with a trust or authority overlay that licenses the opacity to operate. In the equity market the overlay is unusually dense, and it is worth counting its layers. The investment banks that underwrite the offering certify, by their fee and their name, that the price is sound; that is the first layer. The Big Four audit firms whose opinions sit atop every set of accounts are the second. Generally accepted accounting principles, which bless the recognition of circular revenue and the capitalisation of acquisition cost as fair presentation, are the third. The sell-side analysts whose consensus is the public scoreboard are the fourth. The index providers, whose inclusion rules convert size into a compelled bid, are the fifth, and the most quietly powerful, because inclusion is treated as a neutral, mechanical fact rather than the largest discretionary purchase decision in finance. The venture-capital marks that manufacture the private ‘valuations’ the public is later sold are the sixth. And the implicit understanding that a large enough fall in asset prices will be met by the central bank (the put the monetary essay described) stands behind all of them as the seventh. Each layer signals to the saver that someone responsible has already decided the price is reasonable. Usually no one has. The credibility is the product, and the analysis is the thing it is sold in place of.

Functional opacity supplies the material the credibility waves through. A valuation of ninety-four times revenue is rendered legible by a narrative no spreadsheet could support. Adjusted, non-standard earnings metrics convert losses into ‘profitability’ by excluding the costs management would rather not count. The circular financing of the AI economy hides, inside a thicket of disclosed but unmappable mutual deals, how much reported growth is the same capital lapping the same circuit. The private mark is an opinion wearing the clothes of a transaction; the captive bid is a forced purchase wearing the clothes of a free market. The saver does not learn they are inside the combination until the story breaks and the price reverts to the cash the business actually throws off, and by then the discovery is the loss, and it is their retirement that holds it.

Who Holds It When the Story Ends

The cost of this market, when its inflow pauses, falls in a pattern the series has traced through every other structure, except that here it falls on almost everyone at once, because almost everyone has been routed into it. The early holders are gone by then: the venture funds that sold into the IPO, the executives who exercised into the buyback-lifted price, the sponsors paid at the merger. The late buyer who bought the story at the top holds the markdown. But the deepest exposure is the one no one chose. The worker auto-enrolled into an index fund owns, in proportion to its size, the most inflated company in the index, bought with their contributions at the valuation the story justified, by a bid required to buy it. The saver who did everything right (diversified, low-cost, passive, exactly as a generation was correctly taught) has been made, without their knowledge or consent, the marginal funder of the bubble, because the low-cost passive vehicle is precisely the price-insensitive buyer the structure needs. The prudent and the reckless hold the same position. There was no way to opt out that was not also opting out of retirement.

The Everything Bubble

The equity market is not a fraud. It is the greatest engine of capital formation ever built, it has made ordinary people part-owners of the productive economy, and the index fund is a genuine gift to the saver. The achievement is enormous and should not be lost in the criticism of what has been built on top of it.

But the most-rewarded designs in the market have become machines for manufacturing price and inflow ahead of production, and they have reached their most florid form at the very moment the savings of an entire generation have been automatically routed into them. A company that loses money at every line is floated at ninety-four times revenue, and the index is compelled to buy it. Software is valued on stories that revalue by three-quarters when the mood turns. The defining boom of the age runs partly on its own capital travelling in a circle and being counted as growth at every lap. The buying that holds it all up is the forced, price-blind contribution of people who were correctly told that passive, diversified, low-cost investing was the wise course, and were not told that it had made them the last buyer in the chain. On all three axes the verdict is the harshest in the series, and on longevity the assumption that the inflow never pauses, having failed repeatedly in the last five years, is now inflated to a scale the only honest words for which are bubble and, when it goes, implosion.

The authority overlay and the functional opacity have, together, carried the market past every quarter in which the gap between its prices and its production could have been settled and was not. But this is the last floor. There is no deeper bedrock to appeal to, no productive economy underneath the productive economy, no further backstop behind the one that backs everything else. When the story that prices the market finally fails to find its next believer, it will not fail for one company. It will fail under everything that was priced against the assumption that the market beneath it was real: the pensions, the index funds, the bank capital, the sovereign’s tax base, the savings that hold the accumulated work of a generation.

This is what ‘everything bubble’ actually means: not hyperbole, but the literal observation that every structure this series has examined is now leaning on the price of this one, and the price of this one is a story being told to a buyer who is required to believe it. Whether that combination is removable (in this market, or in any Ponzi-like structure of any kind) is the question this essay does not answer. It leaves the question with the reader, who owns the index, holds the retirement account, and is, whether they ever agreed to it or not, the last buyer standing under the most expensive story ever told.

Conclusion · The PUNZI Papers Essay Nine

This completes the present essay of The PUNZI Papers, a series of essays explaining the position behind our project, PUNZI WARS, and the potential it points to.

In the next essay: The Debt-Ponzi We All Find Ourselves Trapped In.

Visit us: https://punzi.xyz Follow us on X: @PUNZIwars

Advisory: PUNZI WARS is a game and an on-chain research experiment. It is not an investment. Participate at your own risk.

Sources

1. SpaceX (Space Exploration Technologies Corp.), Registration Statement on Form S-1, US Securities and Exchange Commission (EDGAR CIK 0001181412), 2026; IPO pricing and Nasdaq listing under ticker SPCX, June 2026

2. SpaceX–xAI merger announcement and Elon Musk blog post / memo (February 2026), source of the ‘sentient sun … extend the light of consciousness to the stars’ mission language, as reported across financial and technology press

3. Aswath Damodaran (New York University, Stern School of Business), valuation of SpaceX at roughly $1.3 trillion against the $1.75 trillion IPO price, and characterisation of the prospectus’s $28.5 trillion addressable-market claim as a figure that ‘borders on fantasy’, with the observation that the filing read as though it might have been written by the company’s own chatbot (June 2026 commentary)

4. Morningstar equity research, ‘Why We Think the SpaceX IPO Is Overvalued’ and ‘Financials Look Reckless: Lifting xAI’s Hood in the SpaceX IPO’ (June 2026), fair value at roughly half the IPO valuation

5. Matt Levine, ‘The SpaceX Squeeze’, Bloomberg Opinion (Money Stuff), June 2026, on the index-fund forced-buying dynamic of the SPCX flotation

6. S&P Dow Jones Indices, S&P 500 Buybacks (quarterly releases, 2018–2025)

7. US Securities and Exchange Commission, Rule 10b-18 under the Securities Exchange Act of 1934 (adopted 1982)

8. William Lazonick, ‘Profits Without Prosperity’, Harvard Business Review (September 2014)

9. US Department of the Treasury, Payroll Support Program disclosures under the CARES Act and subsequent extensions (2020–2021)

10. Meritech Capital and Bessemer Cloud Index, public SaaS revenue-multiple time series (2020–2023)

11. Reporting on AI infrastructure commitments (OpenAI agreements with Oracle, AMD, Amazon Web Services; Nvidia’s stake in CoreWeave; Nvidia–OpenAI investment), 2025, and analyses of circular-financing and round-tripping risk in the AI sector

12. Apple Inc., Annual Reports on Form 10-K (fiscal years 2013 through 2024)