On Friday morning, Nasdaq rang two opening bells at once — one in New York, one in Texas, a first in the exchange’s history. Elon Musk rang his from Starbase. By the time the order book crossed, SpaceX was a public company valued at roughly $1.8 trillion, the seventh most valuable corporation in America on its first print, and Musk was — depending on where the tape settled — the first trillionaire in recorded history.

The financial press spent the day asking whether the stock would pop. That was never the interesting question. The pop was engineered weeks in advance, and we’ll get to how. The interesting question is the one nobody asked at the ceremony: whose money was on the other side of the trade?

Follow it far enough and you arrive at a pension contribution leaving a teacher’s paycheck, routed through an index fund that was never asked for an opinion. That’s the story. Not the rocket, not the trillionaire — the plumbing.

The chain ran dry

To understand why this IPO happened now, at this size, in this shape, start with what’s upstream.

Venture capital and growth equity have been stuck at the exit for years. Distributions back to limited partners — the DPI that pension funds, endowments and sovereigns actually live on — collapsed in 2023 to their lowest level in nearly 14 years, according to PitchBook, and the drought has barely eased since: PitchBook’s Q2 2025 benchmarks put the 12-month distribution yield at 10.9%, against a decade average near 19.6%. The money went in during 2018–2021; it came back as paper markups, not cash. M&A froze under antitrust pressure and interest rates. The IPO window stayed shut for three years. The result is a backlog of unicorns carrying valuations no one can monetize.

Meanwhile, the AI buildout’s appetite for capital escaped the scale of every private pool that exists. SpaceX’s own prospectus is the cleanest exhibit: capital expenditures hit $10.1 billion in the first quarter alone — more than double a year earlier — and $7.7 billion of it went not to rockets but to AI infrastructure. That’s a roughly $40 billion annual run rate, growing, at a single company. Private credit is already saturated with data-center paper. Banks won’t fund burn at that scale against assets that depreciate faster than the loan amortizes.

You can watch the private market run out of marginal buyers in SpaceX’s own tender history: a $400 billion valuation in mid-2025, $800 billion by December, $1.77 trillion at Friday’s pricing. When a price has to double twice in eleven months to find liquidity, the missing buyer isn’t another fund. It’s everyone else.

There is exactly one pool of capital deeper than all of the above, and it had not yet been tapped: the indexed savings of the public. The trillions sitting in 401(k)s, pension funds and passive ETFs that buy whatever the index says, at whatever the price is, every payday, on autopilot.

Friday was the day the pipe was connected.

The index was in the term sheet

Here is what makes SpaceX’s debut structurally different from every large IPO before it: the demand wasn’t found. It was manufactured, and the manufacturing is documented.

Start with the exchange itself. Reuters reported in March that rapid inclusion in the Nasdaq-100 was a key demand from SpaceX in choosing where to list — and Nasdaq, competing with NYSE for the largest listing in history, proposed a new “Fast Entry” rule that compresses index addition for mega-IPOs to under 30 days. Read that again: the rules governing which stocks trillions of passive dollars must buy were rewritten under commercial pressure from a single issuer, as a condition of winning its business. The index stopped being a measurement and became a clause in the deal.

Then the float. SpaceX sold 555.6 million shares — roughly 4% of the company, working back from the $1.77 trillion valuation at $135 per share, which implies about 13.1 billion shares outstanding. The offering was more than four times oversubscribed, with demand reported around $250 billion against $75 billion of stock. The price was fixed at $135, take it or leave it — no book-building range, no upward revision. Holding the price down while demand ran 4x guarantees a first-day pop, and the pop is not a side effect. It’s marketing: it feeds the narrative for the index inclusion, and it ensures that everyone who got cut back in allocation — institutions received a fraction of what they ordered — comes back to buy in the open market.

And then the lockup, which is where the design gets elegant. SpaceX didn’t use the standard 180-day cliff. It built a rolling release: insiders can sell up to 20% of their holdings after Q2 results, with the sellable fraction stepping up at intervals until full release at 180 days. Musk himself signed for 366 days. Now hold that against the index mechanics: indices weight by free float, so every unlock that enlarges the float forces index funds to buy more — mechanically, on schedule, precisely as insiders sell. The passive saver is contractually positioned as the counterparty to every insider exit, by construction.

None of this is illegal. That’s the point worth sitting with. The entire architecture — the rewritten inclusion rule, the starved float, the fixed price, the staggered unlocks — is disclosed, filed, and compliant. An exploit that breaks rules is a crime. An exploit that rewrites them is called strategy, and the difference between the two is the size of the actor. Musk has run this play before: when S&P Dow Jones announced in November 2020 that Tesla would join the S&P 500, the stock jumped nearly 40% in the two weeks that followed, per CNBC. The forced purchase by index funds — what S&P DJI itself called one of the largest funding trades in the index’s history — was estimated at $51 billion at the announcement; by inclusion day, because the price ran in anticipation, the figure had swelled to roughly $85 billion. That $34 billion difference is the mechanism in its purest form: the index funds paid more because everyone knew they would have to pay. SpaceX is the same trade, executed at ten times the scale, with the exchange’s rulebook amended in advance.

The double pipe

The IPO is only the visible half of how public savings fund the AI buildout. The other half runs through debt, and it was Michael Burry — between his depreciation warnings — who pointed at where to look: the special-purpose vehicles.

A layer of intermediary entities has grown up between Nvidia and the companies that actually use its chips. Some are “neoclouds” — CoreWeave, Lambda, Nebius — whose entire business is to buy GPUs, hold them on their own balance sheets, and rent them out under take-or-pay contracts. CoreWeave raised a $7.5 billion debt facility in May 2024, led by Blackstone and Magnetar, secured against its Nvidia GPU fleet — at the time, the largest private debt financing on record. In March 2026 it went further, closing an $8.5 billion facility that it announced as the first investment-grade rated GPU-backed financing in history, anchored by Blackstone Credit & Insurance with insurance investors participating. Read that milestone carefully: depreciating silicon with a two-to-three-year competitive life is now rated collateral that insurers hold against policies.

Some intermediaries are pure SPVs built to keep the hardware off everyone’s books. Meta’s Hyperion data-center campus in Louisiana is owned by a $27 billion joint venture in which Blue Owl Capital’s funds hold 80% and Meta just 20% — Meta leases its own facilities back from the vehicle on four-year initial terms, the debt was placed with PIMCO and other bond investors through a private offering, and Meta collected a $3 billion one-time distribution from the vehicle at formation. A multi-decade asset, financed off balance sheet, rented back on four-year paper.

The purest specimen is the financing built for xAI’s Colossus 2 in Memphis: a $20 billion special-purpose vehicle anchored by Valor Equity Partners — roughly $7.5 billion of equity and $12.5 billion of debt — that buys Nvidia GPUs and leases them to xAI over five years, with the debt collateralized by the GPUs themselves rather than xAI’s corporate assets, per Bloomberg’s reporting. Nvidia itself invested up to $2 billion in the vehicle’s equity, with Apollo and Diameter in the debt. Nvidia financing the entity that buys Nvidia chips to rent to an Nvidia customer. And note where that structure sits today: xAI merged into SpaceX in February. The SPV is now plumbing underneath the company that just rang two opening bells.

Who lends to these entities? Private credit — Blackstone, Apollo, Blue Owl, Ares. And who are private credit’s limited partners? Pension funds, insurers, sovereign wealth. The chain is short and unbroken: a retirement contribution flows into a private credit fund, which lends against GPUs sitting in a warehouse — sometimes literally waiting for a data center with enough electrical power to exist — depreciating toward obsolescence on Nvidia’s own product schedule.

The function of the intermediary layer is risk isolation, and it works beautifully for everyone upstream. Nvidia recognized its revenue at shipment. The hyperscaler holds a cancellable lease, not a stranded asset. The SPV holds the iron, the debt, and the residual risk — and the SPV’s capital comes from the most passive, least informed money in the system. The risk wasn’t eliminated. It was transported down the chain until it found the holder who can neither analyze it nor sell it.

Which means the public saver is now plumbed into the AI cycle twice, through both pipes at once. On the equity side, index funds buy SPCX because the index — amended for the occasion — says so, and the same mechanism stands ready for the next two listings. On the debt side, pension money funds the depreciating hardware that no corporate balance sheet wanted to own. If the cycle cracks, the same pool absorbs the equity writedown and the collateral default. The ant signed neither.

Burry’s historical reference was Enron, and it’s more precise than it sounds. Enron’s innovation was never inventing losses — it was inventing places to park bad assets so the visible balance sheet stayed clean while the risk migrated to whoever couldn’t see it. The 2026 version is larger, fully disclosed in footnotes nobody reads, and legal.

The IPO, inverted

Step back far enough and Friday marks the completion of a reversal twenty years in the making.

The public market used to be where growth happened. Amazon went public at a $440 million valuation and did its compounding in front of everyone; the rules — securities law, index criteria, the standard lockup — were all written for that world, where companies arrived small and earned their index weight over years of public price discovery. Then private capital got deep enough to capture the entire growth curve: SpaceX spent 24 years and every doubling from zero to $1.77 trillion inside accredited-investor markets, and arrived at the exchange fully formed, too large for the rulebook — so the rulebook was adjusted, in about a month, under competition between exchanges.

The IPO has inverted from financing to distribution. The public is no longer buying Amazon’s growth at $440 million; it’s buying the exit of those who captured the growth privately. And Friday’s $75 billion isn’t expansion capital in any traditional sense — it’s plugging a capex burn that private markets can no longer carry.

Now look at the queue. OpenAI filed confidentially for its listing on Monday. Anthropic filed the week before, reportedly at a valuation near $965 billion. Bloomberg calculates the three companies together could add $3.6 trillion of market value to U.S. exchanges. The Fast Entry rule exists now. The staggered-lockup template exists now. The precedent that an issuer of sufficient size can negotiate with the index itself — that exists now too. SpaceX wasn’t an IPO. It was a dress rehearsal, performed twice more before year-end.

The question that was asked wrong

The morning of the debut, the natural question was whether this would be a black Friday — whether the biggest IPO in history might break on the launch pad.

It was the wrong question, and the structure explains why: Friday was designed to be incapable of going black. A 4% float, four times oversubscribed, with a fixed price 20%-plus below indicated demand and an index amendment waiting — that day cannot fall. Every variable was controlled.

The right question is sequential. Private wealth funded this cycle and ran dry. Private credit funded the next layer and is saturating. Friday, the last and deepest pool — the involuntary, automatic, indexed savings of the public — was formally connected to the machine, and two more connections are scheduled. Behind that pool there is no other pool.

So the black Friday, if it comes, won’t be a debut that breaks. It will be an ordinary day when the unlocks are selling on schedule, the index funds are buying on schedule, and the only people who never chose any of it discover they were the bid all along.

Update — June 12, 4:00 PM ET. The tape settled, and it settled lower than the morning advertised: SPCX’s opening auction printed at $150, an 11% pop, below the $160–175 range the indications had been showing for hours. What happened next is the part worth recording. Roughly a third of the entire 555.6-million-share float changed hands in the first ten minutes — the flippers delivering inventory to cut-back institutions and retail in one compressed burst. Then volume collapsed, and the price climbed anyway: $176.52 at the day’s high, before fading into the close to settle at $160.95 — up 19.2%, almost exactly where the morning’s pre-open indications had pointed. By the closing bell, roughly 504 million shares had traded: 90% of the entire float changing hands in a single session. Day one was, in effect, a second distribution — yesterday’s allocations migrating to their real holders. The shares had migrated to hands that don’t sell before index inclusion — Sequoia’s Shaun Maguire, an early investor, told CNBC he’d compare the company to “Nvidia three years ago” and intends to hold his shares forever. Market cap crossed $2 trillion within minutes of the open, and at these prices Musk’s 41% stake alone is worth roughly $900 billion: the trillionaire arithmetic closed before lunch. On JPMorgan’s livestream, Musk said the company has been cash-flow positive since around 2015 and that the raise funds “a significant growth phase” — over 100,000 satellites and AI data centers in orbit. Note what that means: a company that says it doesn’t need the money just executed the largest capital raise in market history, into the only pool deep enough to supply it. And the Fast Entry clock is now running: Nasdaq-100 inclusion lands within 15 trading days, which puts the forced index bid on the calendar for early July — the first scheduled purchase by the savers who never placed an order. The first clean price in SpaceX’s history arrived at today’s close. The first honest one arrives when the lockups do.