Priced for rescue

On June 4, 2026, S&P Dow Jones Indices finished a consultation and decided to change nothing. It reaffirmed the rules that govern which companies are allowed into the S&P 500, and in doing so it shut the door, for now, on the largest initial public offering ever attempted. SpaceX is raising roughly $75 billion at a valuation near $1.75 trillion, against a $4.94 billion net loss in 2025. By S&P's existing rules it is ineligible: it has not been public for the required twelve months, and it is not profitable. S&P declined to waive either requirement, stating that exceptions should not be granted "solely based on market capitalization."

That sounds like a dry governance footnote. It is actually a live readout of the machine this article is about, caught in the act of operating.

Two doors

There is a pipe running through the modern financial system, and the AI-infrastructure bet is built on both ends of it.

The back door is the one people argue about: when a large, interconnected bet goes wrong, the federal government has a documented habit of stepping in to absorb the loss. That is the part that makes the bet rational even if the bet itself is bad.

The front door is the one almost nobody connects to it: the machinery that loads the public's money into these bets in the first place, automatically, before anyone has decided whether they want the exposure. That machinery is index construction, and the SpaceX decision is what it looks like when someone is standing at that door deciding whether to widen it.

Walk through both doors in order, and the logic of a trillion-dollar capital program that does not pencil out on its own terms becomes clear.

The front door: how the public gets loaded in

Most people's retirement money is not actively managed. It sits in index funds, the kind inside a typical 401(k) or target-date fund, which are required to hold whatever is in the index they track, in the proportion the index sets. Nobody picks the stocks. A committee picks the index rules, and the fund mechanically buys whatever qualifies. When a company enters a major index, every fund tracking that index has to buy it, on the same day, regardless of what any individual saver thinks of the company.

That is the mechanism by which the public has already been loaded into the AI-infrastructure bet without being asked. As the five-siphons piece documented, the combined weight of Meta, Alphabet, Amazon, and Oracle in the main US investment-grade bond index nearly doubled in a single year, from 2.2% to 4.1%, as those companies issued roughly $121 billion in new debt against a 2020 to 2024 average near $28 billion. A 401(k) holder in a broad stock fund now carries roughly a quarter to a third of their US equity in the same handful of names. None of them chose it. They were rebalanced into it by the math of index construction.

The scale of that is not hypothetical. On 2 June 2026, Vanguard's S&P 500 fund (VOO) became the first exchange-traded fund ever to cross $1 trillion in assets, the largest fund of its kind in the world, having pulled in some $69 billion in 2026 alone (Bloomberg; Morningstar; ETF.com). Every dollar of that is allocated by market value, so the largest companies, the AI-leveraged cluster at the top of the index, take the largest share of every new inflow. It is a feedback loop, money flows into the index, the index buys more of whatever is already biggest, and the biggest names are exactly the ones whose capex-to-revenue Goldman calls historically unprecedented.

The SpaceX fight is the same mechanism, one level upstream, and it shows the gate being actively loosened. To keep passive money from being force-fed into brand-new, unproven companies, the major indexes historically required a few things: a seasoning period (you have to have been public for a while, typically twelve months), profitability, and a minimum public float (enough shares actually trade freely, measured by what the industry calls the investable weight factor). Those rules are the front door's lock. Here is what each index provider just decided to do with that lock for a $4.94 billion-loss company:

Index provider	What they decided for mega-IPOs like SpaceX
S&P 500 (S&P Dow Jones)	Held the line (4 June 2026). Kept the 12-month seasoning period, the profitability requirement, and the float rule for the S&P 500, and declined the proposed 6-month seasoning waiver. No exception for size. (It did ease additions to its broad total-market indexes from 8 June, but the flagship 500 held.)
Nasdaq-100	Loosened (effective 1 May 2026). New "Fast Entry" rule evaluates a top-40-by-market-cap listing on its 7th trading day and adds it shortly after; the 10% minimum-float requirement was eliminated outright.
FTSE Russell	Loosened. Companies large enough for the Russell Top 500 are added after the 5th trading day post-listing, instead of waiting for quarterly reconstitution. SpaceX qualifies for the Russell US indexes and the FTSE Global Equity series.
CRSP (the indexes Vanguard tracks, $3T+)	Admits low-float megacaps on a float-adjusted basis within days, an estimated $15-25B of mechanical buying on its own.
MSCI (global benchmarks, MSCI World / ACWI)	Open by a rule in place since 2007. Large IPOs above its size thresholds (about $26B free-float-adjusted) enter after 10 trading days; SpaceX clears that and lands in MSCI World and ACWI. An estimated $12B of MSCI World tracker exposure, though the small initial float dampens the first-day weight.

Four of the five major index families leave the front door open to a company like SpaceX on the strength of its size alone. Nasdaq and FTSE Russell widened theirs in 2026, MSCI's has stood open to large IPOs since 2007, and CRSP admits on float-adjusted size. S&P is the lone holdout, and its hold is stronger than a delay: because the S&P 500 also requires profitability, a money-losing SpaceX does not qualify at all until it turns a profit, however long it has been public. Everywhere else, the direction is unambiguous. Where the rules did change in 2026, at Nasdaq and Russell, they moved one way only: toward letting size in sooner, with fewer of the checks that existed precisely to prevent it.

The scale is not abstract. Across the indexes that loosened, analysts estimate the rules will force on the order of $15 to $30 billion of mechanical buying into SpaceX in the months around inclusion, with passive funds potentially absorbing close to a fifth of its freely traded float. None of that buying reflects a view that the company is worth $1.75 trillion. It is the index doing what it does. The financial press named the dynamic without flinching: Fortune wrote that SpaceX "is a massive money loser" that ordinary investors will "be forced to buy anyway," and the Motley Fool called the passive holder, simply, "the exit liquidity."

That is the front door. Now the back.

The back door: the documented habit of socializing losses

The reason the front door matters is what sits behind the back one. When a bet large enough to threaten systemically important institutions goes wrong, the federal government has stepped in to absorb the loss five times in eighteen years. This is not speculation. It is a record.

The five episodes, briefly:

• 2008. The Troubled Asset Relief Program authorized $700 billion to recapitalize banks. The insurer AIG, which had written enormous derivative guarantees it could not honor, received $182 billion in separate lifelines. General Motors and Chrysler got their own auto-industry rescues.
• 2019. The plumbing of the financial system seized. The overnight "repo" market, where banks borrow cash against safe collateral for a single night, spiked from roughly 2.2% to 10% as demand overwhelmed dealers. The Federal Reserve restarted balance-sheet expansion and, in July 2021, made the Standing Repo Facility permanent, a $500 billion on-demand backstop that had not existed as standing infrastructure before the 2019 scare required it.
• 2020. The CARES Act authorized $2.2 trillion, with roughly $500 billion for airlines and large employers; the Paycheck Protection Program added about $800 billion; and the Federal Reserve, for the first time in its history, bought corporate debt directly.
• 2023. When Silicon Valley Bank and Signature failed, regulators invoked the FDIC's "systemic risk exception" to protect uninsured depositors above the $250,000 cap, effectively making the technology and venture-capital ecosystem whole with public exposure.
• 2024 to 2026. Regional banks with heavy commercial-real-estate exposure have received quiet capital infusions and forbearance, and the Standing Repo Facility has been expanded and made operationally permanent.

The pattern is consistent. When the failure mode threatens systemically important institutions or major employers, the government socializes the loss. The post-2008 rules that were supposed to end this, the "living wills" and resolution regimes, did not. The 2023 SVB rescue in particular established that the policy is exception-based, not rule-based, which is another way of saying the exception is the rule.

Why this makes the AI bet rational

Now put the two doors together, and the trillion-dollar capital program that does not close on its own math starts to make sense.

The AI-infrastructure bet's upside flows to equity holders: the founders, executives, and shareholders of roughly a dozen companies. The downside, if it triggers, does not land on them first. It propagates outward, through the pension funds that hold the bondholder debt, through the bank capital concentrated in commercial real estate, through regional banks, through municipal tax bases, and through the index funds that quietly rebalanced ordinary savers into the position. Every one of those failure modes individually meets the "systemically important" threshold that has triggered a federal rescue before. A coordinated AI-capex unwind would meet it overwhelmingly.

Under that calculation the capex is not irrational, even if the skeptics are right that AI will not earn enough to justify it. The bet is not, at its core, a bet on AI's commercial success. It is a high-confidence bet on the federal government's documented pattern of socializing losses when those losses concentrate in systemically important institutions. If AI delivers, the equity holders capture the productivity gain. If it does not, the precedent says the equity holders are still made whole through the rescue mechanism, and the realized cost lands on ratepayers, communities, taxpayers, passive investors, and other businesses.

Private gains, socialized losses, priced into the return math before the bet is placed.

Where SpaceX fits

SpaceX is not a hyperscaler, and this is not a claim that it will be bailed out. It is a claim about the machinery. A company losing nearly $5 billion a year, valued at $1.75 trillion, is precisely the kind of bet that the front door exists to keep out of the public's automatic holdings until it has proven itself. Four of the five major index families let size carry it past the line anyway, some through rules they widened in 2026, MSCI through one it has had for years. The one that refused, S&P, refused on principle as much as on timing: its profitability rule keeps a money-loser out until it actually makes money, which a fast-entry waiver would have erased.

That is the front-end of the same pipe whose back-end is the bailout record above. The public gets loaded in at one door, increasingly fast and with fewer checks, and is positioned as the counterparty of last resort at the other. The gains concentrate at the equity layer in between. The SpaceX decision is worth watching not because SpaceX is the risk, but because it is the clearest recent picture of who controls the front door and which way they are moving it.

What to watch

Three signals will show whether the front door keeps widening:

• Whether S&P's holdout survives its own timeline, or whether the Q4 2026 to Q1 2027 window quietly becomes a fast-track after all.
• Whether the hyperscaler weight in the major bond and equity indexes keeps climbing past the levels the five-siphons piece flagged, pulling more passive money in by construction.
• Whether the first crack in AI-infrastructure credit, an unexplained downward revision to capex guidance, a widening in hyperscaler bond spreads, triggers the back door, and on what terms.

The official story is that the AI boom is financed by venture capital and corporate cash flow. The unofficial story is that it is increasingly financed by the public, loaded in through the front door by index construction and held in place at the back by a rescue pattern that has run five times in eighteen years. The SpaceX decision is a reminder that someone is standing at the front door right now, deciding how wide to open it.

Source-canon corroboration

The article's claims map to The Mild Take's curated resource canon as follows.

Claim type	Canon resource	Methodology angle
Index-inclusion rules; SpaceX eligibility	Primary methodology documents (S&P Dow Jones consultation outcome; Nasdaq Global Indexes NDX methodology + May 2026 FAQ; LSEG / FTSE Russell IPO Fast Entry consultation + index notices; CRSP methodology; MSCI megacap-IPO methodology and research); Ground News spread-check	The index providers' own published rule changes are primary and beat any press framing; US financial press (Bloomberg, CNBC, Fortune) carries the ownership-tilt flag and is triangulated against them
Passive-investor concentration; systemic exposure	IMF (Global Financial Stability Report on AI-infra debt concentration), FiveThirtyEight data	IMF GFSR is the site's preferred independent angle on index-concentration and systemic risk
Federal bailout precedent (TARP, repo/SRF, CARES, SVB, CRE)	Primary US Treasury / Federal Reserve / FDIC releases; IMF GFSR macroprudential analysis	The bailout record is in primary releases; the "exception-based, not rule-based" read is the structural angle IMF macroprudential work supports
State capture; financial-regulation carve-outs	Transparency International, ICIJ, Integrity Index	The canon's elite-capture and Congressional-lobbying layer on Dodd-Frank carve-outs and financial-regulation influence

Where canon runs thin: index-methodology decisions have no multilateral aggregator, so the index providers' own consultation documents are the primary source. SpaceX's private financials are disclosed through the IPO process and reported widely; the $4.94B 2025 loss and ~$1.75T valuation are the figures common across the coverage and should firm up against the final prospectus.