SpaceX (SPCX) opened trading on June 12 at $150 and closed its first day at $160.95. Four days later, on June 16, it reached $225.64. Then options trading began on June 17. As of this week, SPCX is trading at approximately $164 — down 27 percent from its peak, back near where it finished on listing day.
Thirteen days. A 67 percent run. A 27 percent correction. And 96 percent of shares are still locked until December.
On June 8, OpenAI filed its confidential S-1 with the SEC, targeting a public listing as early as September 2026. The company’s most recent private valuation is $852 billion, set during a $122 billion funding round closed in March 2026. Goldman Sachs, Morgan Stanley, and JPMorgan are leading the offering. The target at listing: above $1 trillion.
Inside that filing, one disclosure stands out above all others. For the equity stake held by Sam Altman — the CEO of what may become the most valuable company in history — the S-1 lists the figure as TBD.
Nobody building a company toward a $1 trillion public valuation has ever entered their S-1 with the CEO’s ownership stake unresolved. The governance structure is unprecedented. The float will be thin. The listing timeline is September. And behind OpenAI, Databricks is preparing to file its own S-1 at a valuation between $134 billion and $175 billion — where its only meaningful public comparable trades at roughly one-fifth that multiple.
The FOMO contagion playbook that migrated from crypto into equity markets via SpaceX is not a one-time event. It is a queue.
What SpaceX Proved in Thirteen Days
In the first analysis in this series, published June 21, we documented the structural parallel between crypto listing mechanics and the SpaceX IPO: a company with genuinely excellent fundamentals, listed at a price that Morningstar placed $1 trillion above fair value and that Damodaran placed $400 billion above his DCF estimate. Former Nasdaq CEO Robert Greifeld said the stock was “not trading on fundamentals.” Bloomberg’s meme-stock columnist applied the meme-stock framework before the first week was out.
In the second analysis, published June 23, we quantified the mechanism. Four percent of total shares were available to trade. Ninety-six percent were locked until December 2026, with a partial release in August. That float structure meant there was no borrowable supply for short sellers. Without a short mechanism, downward price discovery was disabled. On June 17 — five days after listing — options trading began on SPCX. For the first time, bearish positioning was structurally possible. The price has not recovered since.
Gary Black of The Future Fund, who manages portfolios for institutional clients and had avoided commenting on SPCX, broke his silence in the days after listing: “I have resisted commenting on SPCX as it acts more like a meme stock than one driven by fundamentals. Investors should revisit after the August lockup expiry.” Black’s phrasing was deliberate. He was not saying SpaceX is a bad company. He was saying the stock’s behaviour — its price action, its disconnection from intrinsic analysis — is meme-stock behaviour regardless of underlying quality.
The data now confirms the observation. A $225.64 peak reached before any short mechanism existed. A return to $164 the moment options gave the market a way to express a bearish view. The trajectory from listing day close ($161) to peak ($225) to current ($164) is a near-perfect illustration of what supply suppression does to price discovery: it allows narrative to run unchecked until mechanics change.
December is when the real test arrives. When 96 percent of shares become tradeable, the market will price SpaceX against the full supply for the first time. If the stock is below $135 in December — the IPO price — every retail buyer who purchased during the June FOMO run will have lost money on a company that remains one of the most technically impressive in the world. The company’s quality and the stock’s pricing are not the same question.
The Crypto Mechanism, Applied
The academic framework for what is happening here was established by Xu and Livshits in their 2019 paper published at USENIX Security (arXiv: 1811.10109). Studying approximately one hundred Telegram pump-and-dump channels across roughly 412 coordinated events, they found that price peaks within eighteen seconds of a coordinated announcement, then falls below open within three and a half minutes as insiders exit into the retail FOMO wave. The mechanism has three components: insider pre-purchase into a constrained float, coordinated signal release to retail buyers, and rapid insider exit before supply normalises.
Equity market IPOs do not involve criminal coordination. The insiders are not dumping into a pump they engineered. But the structural conditions they create — thin float, locked supply, retail access to a narrative-driven security at peak excitement — generate the same price dynamics across a longer timeframe. Where crypto peaks in eighteen seconds, SpaceX peaked in four days. Where crypto reverts in three and a half minutes, SpaceX has taken thirteen days to fall 27 percent with the largest correction still ahead in December.
We have documented this pattern in crypto markets specifically. A Binance listing of a low-float token called Rayls (RLS) — which we covered when it crashed 80 percent — demonstrated the same mechanics at crypto speed. Small circulating supply relative to total tokens, a listing moment that concentrates buyer FOMO, a collapse when the float dynamics became visible. The difference between Rayls and SpaceX is not structural. It is the speed of the cycle and the reputation of the underlying asset.
The FOMO contagion thesis holds that this structural playbook has migrated from crypto to equity markets — that the psychological and mechanical conditions crypto normalised over a decade have found their way into how traditional equity listings are structured and priced. SpaceX is the proof of concept. OpenAI is the next test.
OpenAI: A $1 Trillion Company With a CEO Who May Own Nothing
OpenAI confirmed its confidential S-1 filing on June 8. The company targets a public listing in September 2026 at a valuation above $1 trillion. Goldman Sachs, Morgan Stanley, and JPMorgan are the lead underwriters — the same institutional machinery that took SpaceX public three months earlier.
The financial profile that S-1 will disclose is substantial by any measure. Revenue estimates from institutional analysts place OpenAI at approximately $25 billion in annualised revenue — a figure that would make it one of the fastest-growing software companies in history. ChatGPT has over 800 million active users. The enterprise API business services virtually every major technology company running AI infrastructure. OpenAI’s competitive position is real, its revenue is real, and its growth trajectory is documented.
None of that is the problem.
The problem is structural. OpenAI was founded in 2015 as a nonprofit research organisation with an explicit mission: to ensure that artificial general intelligence benefits all of humanity. That foundation is not a historical footnote. It is still part of the legal architecture. The company completed its conversion from a nonprofit to a Public Benefit Corporation in late 2025, but as part of the settlement with the nonprofit foundation, that foundation retained a stake valued at approximately $130 billion — and, more importantly, retained governance rights over the company that do not transfer to public shareholders.
In a conventional technology IPO, a dual-class structure concentrates voting power with the founders. Mark Zuckerberg controls Meta through Class B shares that carry ten times the voting weight of public shares. Alphabet operates on the same model. In both cases, public investors know exactly who they are deferring to and what economic interest that person holds. The alignment is imperfect but legible.
OpenAI’s structure is not legible in the same way. The nonprofit Foundation retains control post-listing through mechanisms that public investors rarely encounter at this scale. And the CEO’s equity — the number that would tell you how much Sam Altman benefits from the company succeeding — is listed in the filing as TBD.
Congressional investigators and state attorneys general have been examining whether Altman’s personal investment portfolio creates incentive structures that conflict with OpenAI’s stated mission. That scrutiny was already documented before the S-1 filing. What the S-1 now adds is the extraordinary disclosure that, at the time of filing, the equity structure for the world’s most watched CEO in the world’s most anticipated technology listing had not been resolved.
We examined OpenAI’s governance structure in detail in an earlier analysis of Altman’s conflicts and what the IPO structure reveals. The S-1 filing confirms that the governance concerns were not overstated. Buyers of OpenAI stock at listing will be purchasing a minority economic interest in a Public Benefit Corporation whose core decisions remain with a nonprofit foundation whose beneficiary is not the shareholder.
This adds a dimension to the FOMO contagion thesis that SpaceX did not. With SpaceX, the FOMO was straightforward: an excellent company, priced at peak narrative excitement, on a float that suppressed short selling. With OpenAI, the FOMO will run on a story — the story of AGI, of ChatGPT, of the company that made AI real for a billion people — while the governance structure ensures that public shareholders are structurally subordinate to a mission-driven body whose priorities may not always align with share price.
At $1 trillion, OpenAI would be trading at approximately 40 times its estimated revenue. Microsoft, which integrates OpenAI’s models throughout its product suite and owns a 49 percent stake via multi-year investment agreements, trades at roughly 12 times forward revenue. Apple, with a hardware ecosystem that creates recurring software and services revenue, trades below 10 times. The $1 trillion valuation price embeds not just current AI leadership but permanent AI leadership — a category winner assumption that may or may not survive the next three years of compute competition from Google, Anthropic, Meta, and the growing cohort of open-source model developers.
The float will be thin. A single-digit float percentage on a $1 trillion company implies tens of billions in proceeds while leaving the vast majority of shares locked. The same supply mechanics that concentrated FOMO in SPCX’s 4 percent float will operate in OpenAI’s listing. The narrative is larger. The governance complexity is greater. The listing price will be set at peak excitement — September 2026, when AI remains the dominant investment theme of the decade.
The lockup expiry is when public investors discover what they actually bought.
Databricks: 32x Revenue When the Comp Trades at 7x
Databricks is the third data point in the queue. In December 2025, the company closed a $5 billion Series L funding round at a $134 billion valuation. By June 9, 2026, The Information reported that Databricks was in talks to raise a new round at a valuation between $165 billion and $175 billion. The S-1 filing is expected in Q3 2026, with the listing likely landing in Q4 2026 or early 2027.
The financial profile is genuinely strong. Databricks has a $5.4 billion annualised revenue run rate, growing 55 percent year-over-year. It has more than 650 customers each generating over $1 million in annual revenue. It is the dominant data intelligence platform for enterprises integrating AI into their operations. The company is real, profitable in the right metrics, and growing fast by any reasonable measure.
The valuation problem is not Databricks’ fundamentals. It is the multiple at which private investors priced those fundamentals, and what public markets will make of that multiple when the S-1 lands.
At $175 billion, Databricks would be priced at approximately 32 times its revenue run rate. Its closest public comparable is Snowflake — a data cloud company competing in overlapping enterprise workloads, with substantial revenue and a well-understood growth trajectory. Snowflake trades at roughly 7 times forward sales. Its public market capitalisation sits near $58 billion.
The gap between 32x and 7x is not a small valuation premium. It represents a market that has priced Databricks like a generative AI model lab rather than a software vendor. As one institutional analysis put it: at 32 times run-rate, Databricks would need to sustain exceptional growth — not decelerate even from 65 percent to 45 percent — to justify the number. Any meaningful slowdown in revenue growth would trigger a repricing toward Snowflake’s multiple, which at equivalent revenue would imply a market capitalisation roughly one-fifth of $175 billion.
“A Databricks listing at $175 billion would be among the largest technology IPOs ever — and would force a public revaluation of the entire data platform category.” That phrasing deserves attention. Forcing a public revaluation means Databricks’ listing price would set a new multiple reference for all enterprise data companies. When SpaceX listed at $1.77 trillion, it reset the benchmark for what “space and satellite infrastructure” could command. When it corrects, that benchmark corrects too. The companies that priced off the SpaceX peak will mark themselves down alongside it.
Databricks would do the same thing to enterprise AI infrastructure. List at $175 billion, set a 32x multiple as the visible market reference, watch the entire category price off that number — until the lockup expiry forces the market to confront supply that was never priced in on listing day.
Three Companies, One Playbook
The pattern across SpaceX, OpenAI, and Databricks is not a conspiracy. There is no coordination between these companies or their underwriters to exploit retail buyers. The playbook runs itself, because the incentive structure of private-to-public transitions makes it the rational choice for every participant at every stage.
Private investors — venture firms, sovereign wealth funds, secondary buyers — spent years accumulating positions at valuations that were themselves priced on narrative and growth expectations. Their exit is the IPO. Their optimal outcome is a listing price that clears above their cost basis by the widest possible margin, sustained long enough to distribute their position without crashing the stock. A thin float serves this interest directly: it suppresses the supply that would allow the market to find fair value while their lockup prevents them from selling anyway. The FOMO run is their paper gain. December is when they find out how much they actually made.
The underwriters — Goldman, Morgan Stanley, JPMorgan — price the offering to generate a first-day pop. A first-day pop creates the narrative that the deal was “successful.” It rewards institutional allocatees who received shares at IPO price and flipped them on listing day. It creates press coverage that reinforces the FOMO. The pricing methodology optimises for listing-day performance, not for where the stock trades when full supply enters the market six months later.
Retail buyers — who increasingly participate in IPOs via brokerage platforms, Reddit threads, and financial content creators — enter when the narrative is at its most concentrated. OpenAI will be the most discussed company in the world in the weeks before its September listing. The ChatGPT story, the AGI story, the “most transformative technology since the internet” story will be everywhere. Retail buyers will bid for access to a story, paying a price set by that story’s peak concentration, with no mechanism to short the stock and no ability to borrow shares that do not yet exist in the float.
This is the FOMO contagion thesis in its most distilled form. Not fraud. Not manipulation in the legal sense. A set of structural incentives that produce, reliably, the same outcome: listing price equals narrative peak, thin float suppresses the correction, lockup expiry delivers the reckoning.
In crypto, this cycle ran at extraordinary speed because the market never slept, settlement was instant, and there were no lock-up periods. Xu and Livshits documented the complete cycle — announcement, peak, collapse — in under four minutes per event. In equities, the regulatory architecture slows the cycle dramatically. But it does not change the outcome. SpaceX peaked in four days. It has been correcting for thirteen. OpenAI will peak sometime in September or October. It will correct over a longer arc. The mechanism is the same.
The OpenAI Governance Problem Is a New Variable
SpaceX and Databricks both have straightforward equity structures, even if their valuations are contested. Elon Musk owns SpaceX. Ali Ghodsi founded Databricks. Their equity positions are large, known, and incentive-aligned with the stock price. Retail buyers in both cases are buying into a company where the founder has overwhelming reason to make the stock perform.
OpenAI’s structure is different in a way that matters for anyone buying at the listing price.
The nonprofit Foundation that retains governance control post-IPO has a primary obligation to OpenAI’s stated mission — the safe and beneficial development of artificial general intelligence for the benefit of humanity. That is not an obligation to maximise shareholder returns. It is not a commitment to pursue the most profitable product strategy. It is a mission obligation that may at times conflict with what a conventional public company would do to enhance its stock price.
The examples are not hypothetical. OpenAI has historically published research that arguably benefited its competitors — open publication of safety methodologies, architectural innovations, and evaluation frameworks that allowed others to build better models. A shareholder-maximising board would arguably limit such publication. The nonprofit Foundation’s obligations point in the opposite direction. Public shareholders will have limited ability to influence which outcome prevails.
The CEO equity situation compounds this. When the S-1 was filed, Sam Altman’s ownership stake in OpenAI was unresolved. This is not a disclosure formality. It tells institutional investors that the company has not yet finalised the foundational alignment question of any tech listing: how much does the person running this company make if it succeeds? Without that number, the governance structure that would normally provide some check — CEO equity creating alignment with shareholder interests — is absent from the analysis.
Congressional scrutiny of Altman’s investment portfolio adds another dimension. If Altman holds significant positions in companies that supply OpenAI with compute, data, or distribution — and if the nonprofit Foundation’s governance structure makes it difficult for a standard shareholder vote to remove or restrain him — then the “TBD” equity situation is not merely incomplete. It may be strategic ambiguity that allows Altman to navigate the conflict disclosures without committing to a number that makes the conflicts legible.
None of this will suppress the FOMO when the S-1 becomes public. The ChatGPT narrative is too strong and the AI investment thesis is too dominant in 2026 for governance concerns to materially suppress listing-day demand. But governance concerns compound over time. They become visible at earnings calls when management cannot explain decisions in shareholder-return terms. They become visible when the Foundation vetoes a product strategy that the business logic favoured. They become visible at the first lockup expiry, when institutional holders who read the S-1 carefully decide whether to distribute or hold.
What the December Test Will Show
SPCX in December is not just about SpaceX. It is the first publicly observable test of whether the FOMO contagion thesis makes a prediction that comes true.
The thesis predicts: when 96 percent of shares become tradeable, the market will price SpaceX against full supply for the first time. That price will be determined by fundamental analysis, institutional models, and supply-demand dynamics with a complete float — not by narrative concentration in a 4 percent float. Whether that price is above or below $135 — the IPO price — is the empirical test.
The current trajectory argues for a significant further correction. SPCX is at $164, having fallen 27 percent from its $225.64 peak in thirteen days after options began trading. The August partial lockup expiry, which Gary Black specifically identified as the next structural inflection point, will introduce more shares to the float before December arrives. Each incremental supply event pressures the stock toward fair value, before the December event delivers the full test.
Morningstar’s fair value estimate was $780 billion in early June, before the IPO. Damodaran’s DCF placed fair value at $1.3 trillion. The IPO priced at $1.77 trillion implied market capitalisation. At $164 per share, SPCX currently implies approximately $1.47 trillion — still above both analyst estimates, but on a trajectory toward them.
Investors who bought at $225 in the first week of June need SPCX to recover above that price and stay there through December to break even. The structure suggests that path is unlikely: more supply enters in August, first earnings arrive September 2, and the full lockup expiry in December adds the remaining 96 percent of total shares. Each event is a downward pressure on the narrative-driven pricing that $225 represented.
The Cursor acquisition adds complexity to the analysis. SpaceX committed to a $60 billion all-stock acquisition of the AI coding platform Cursor before the IPO — a pre-commitment made in April 2026 when the deal was structured, before SPCX began trading. We documented in our second analysis that at $225 per share, SpaceX issues approximately 40 percent fewer shares to complete the Cursor acquisition than it would have at $135. The FOMO run was materially beneficial to SpaceX as an acquirer in a way that retail buyers who funded it did not necessarily understand. Their FOMO subsidised the deal.
December will tell the story. It is either the moment SpaceX demonstrates that public market fair value exceeded the pessimists’ estimates, or it is the moment the thesis is confirmed in full: a genuinely excellent company whose listing event was priced at narrative peak and whose stock spent the following six months returning toward a value the market would support without the supply constraint.
The Queue Behind OpenAI
OpenAI in September and Databricks in Q4 2026 or Q1 2027 are not isolated events. They represent the largest concentration of high-anticipation private company listings in a single 18-month window since the 2000 dot-com bubble peak. At that peak, Cisco completed 72 acquisitions using its inflated stock before the supply normalised and the market corrected 80 percent from its high. AOL acquired Time Warner in a $182 billion all-stock deal at the precise moment AOL stock was priced on a growth narrative that public markets would not sustain for another year.
The parallel is not that AI is a bubble — AI is a genuinely transformative technology, as the internet was genuinely transformative. The parallel is structural: a period when the distance between narrative pricing and fundamental analysis is at its largest, when the companies best positioned to exploit that gap are doing so rationally, and when retail participation is at its highest concentration of optimism.
OpenAI at $1 trillion needs to be the permanent winner in a model race where Google, Anthropic, Meta, Mistral, and a growing cohort of open-source developers compete with billions of dollars in compute investment. Databricks at $175 billion needs to sustain growth that justifies a 32x revenue multiple at a time when enterprise AI infrastructure is still consolidating and no category winner has been definitively established. Both companies may be worth those numbers in ten years. But the listing price is not a ten-year price — it is the price at peak narrative, set by a float structure that suppresses the market mechanism that would otherwise find a lower equilibrium.
Anthropic has not announced a listing timeline. Private valuation estimates vary between $60 billion and $100 billion, with the company focused on safety research and enterprise API revenue. Its listing, when it comes, will carry its own narrative concentration — the “safe AI” story, the Amazon partnership, the Claude model family. The same structural playbook applies regardless of which narrative is the vehicle.
The common thread across SPCX, OpenAI, Databricks, and whatever follows is not that these companies are bad. They are not. It is that the listing event — as structured — is never the right price. It is the peak price, made available to retail at the moment of maximum narrative excitement, with the minimum possible float, and the maximum possible lockup protecting insiders while the market absorbs supply it cannot yet price correctly.
What Retail Buyers Are Being Asked to Do
The question that OpenAI’s September listing will ask of retail buyers is the same question SpaceX asked in June: are you pricing the company, or are you pricing the story?
Pricing the company requires a view on revenue trajectory, competitive dynamics, margin structure, governance risk, and lockup mechanics. It requires an assessment of what $1 trillion implies in terms of free cash flow over a twenty-year horizon and whether a company running on GPU compute from Nvidia and Microsoft Azure can sustain the margins that $1 trillion assumes. It requires a view on what the CEO’s undisclosed equity position means for management incentives and the nonprofit Foundation’s influence on product strategy.
Pricing the story requires none of that. It requires only that you believe OpenAI is the company that made AI real, that ChatGPT is the most important product of the last decade, and that the name on the listing is worth more than what anyone who runs a DCF model says it is. That belief is widespread, emotionally grounded, and — in the days before a September 2026 IPO — it will be the dominant input into a market structured to accommodate it.
The institutional buyers who receive IPO allocations at the listing price understand this distinction. They are buying a first-day pop opportunity, not a long-term position. The long-term positions — the ones that reflect a genuine view on fair value — are built in the secondary market, at prices that reflect supply the float structure will not allow on listing day.
Retail buyers, who access the stock at open-market prices on listing day, are buying after the institutional pop has already occurred. They are the market at which institutional allocatees distribute. They are the supply absorbers that the thin float requires — willing buyers who provide exit liquidity for the early holders who cannot yet exit via lockup and for the institutional allocatees who were paid to take the first-day risk.
This is not how it is described in the prospectus. But it is how the mechanics work. SpaceX has provided, in thirteen days, a visible example of exactly how they work.
The Thesis Stands, and It Has More Data Coming
The FOMO contagion argument, which we first advanced in June, is that the psychological and structural mechanics that crypto markets normalised over a decade — the listing event as narrative peak, the thin float as price discovery suppressor, the lockup expiry as the real exit event — have migrated into traditional equity markets at scale.
SpaceX in June provided the anchor case. A company listed at $1.77 trillion against two serious analyst estimates of $780 billion and $1.3 trillion. A 4 percent float that produced a 67 percent run in four days by suppressing bearish positioning. A correction that began the moment options trading gave the market its first short mechanism. A trajectory toward lockup expiry in December that will, for the first time, price SpaceX against the full supply of shares its fundamentals must support.
OpenAI in September adds a company with genuinely unclear governance — a CEO whose equity is TBD, a nonprofit foundation with control rights that subordinate public shareholders to a mission, and a valuation that implies permanent AI leadership in a field where the competitive dynamics change every six months. The float will be thin. The narrative will be enormous. The governance complexity will become visible after the listing excitement fades.
Databricks adds the enterprise software multiple compression story — a company priced at 32 times revenue preparing to list in a market where its most comparable public peer trades at 7 times. The listing will either force the market to accept a new multiple for the category, or the market will decline to accept it and Databricks will mean-revert toward the 7x that public enterprise software has historically supported.
By December 2026, there will be three visible datasets: SPCX with a full float, OpenAI in the first months of trading, and Databricks approaching its own listing. Together, they will either confirm the thesis across multiple examples or falsify it by demonstrating that public markets are willing to sustain narrative-peak pricing once the full supply is available.
The SPCX trajectory through June suggests the former is more likely. The thesis has survived its first thirteen-day test. It has two more datasets queuing up behind it.
