Start with tomorrow. June twelfth, SpaceX lists on the Nasdaq, ticker SPCX, at a hundred and thirty-five dollars a share. The valuation — about one and three-quarter trillion dollars. It raises roughly seventy-five billion. Per the New York Times, that one raise is more than every US IPO of the last two years combined. The largest IPO in history, by a wide margin.
This is YegerPod. I'm Yeger. It's Thursday, June eleventh. Today in AI, a special one — the money. SpaceX, Anthropic, OpenAI, and the question underneath all of it: how are these numbers even possible?
Start with tomorrow. June twelfth, SpaceX lists on the Nasdaq, ticker SPCX, at a hundred and thirty-five dollars a share. The valuation — about one and three-quarter trillion dollars. It raises roughly seventy-five billion. Per the New York Times, that one raise is more than every US IPO of the last two years combined. The largest IPO in history, by a wide margin.
Now the numbers underneath it, because they're the whole point. SpaceX's 2025 revenue — eighteen point seven billion, up thirty-three percent. Its net loss for the year — four point nine billion. Starlink is the only profitable division. And the AI unit folded in — that's xAI, merged into SpaceX back in February at a one-and-a-quarter-trillion mark — burned seven point seven billion dollars in just the first quarter of 2026, on a two-and-a-half-billion operating loss. Those are Bloomberg's figures. So the most valuable IPO ever is a company losing billions, valued at nearly ninety times its revenue.
Same week, the other two. Anthropic filed a confidential S-1 on June first, at a nine hundred and sixty-five billion dollar valuation, on a run-rate around forty-seven billion. OpenAI confirmed its own confidential filing June eighth. Three of the largest capital events in history, stacked into ten days.
So — Bob's question, and it's the right one. How are these amounts possible? Let me give you what analysts are actually pointing at. Two mechanisms, then the macro.
Mechanism one — circular financing. This is the phrase showing up everywhere from Man Group to the Wall Street commentary. The idea: a tight web of deals where the same dollars cycle between a handful of companies and inflate everyone's revenue. Man Group calls it, quote, "a closed, recursive financing loop across hyperscalers and their strategic partners." The same names — Microsoft, Nvidia, Amazon, Google, OpenAI, Anthropic — act at once as suppliers, customers, and investors to each other.
Concretely, the way one analysis lays it out: Microsoft invests in OpenAI; OpenAI spends it on Azure compute, which books as Microsoft revenue, which justifies Microsoft's capital spending — reported around a hundred and ninety-two billion. Alphabet and Amazon are bound to Anthropic the same way — the equity stake and the cloud contract are the same bet. Unwind one, you unwind both. Money moving in a circle can look like growth at every step.
Mechanism two — the gap between the spend and the revenue. J.P. Morgan modeled what the whole AI buildout would have to earn to clear a basic ten-percent return on the capital going in. The answer — about six hundred and fifty billion dollars a year in AI revenue. The current run-rate is roughly twenty-five billion. That's a twenty-six-fold gap. J.P. Morgan's own framing of what closing it requires: thirty-five dollars a month, in perpetuity, from every iPhone user on Earth.
And it's not only the skeptics saying the gap is real. Goldman Sachs' chief economist concluded AI contributed, quote, "basically zero" to US economic growth in 2025, and that — his words — "FOMO, not ROI, is driving hyperscaler capex." The San Francisco Fed, this February: generative AI is useful, but "not the innovation that spurs broad-based reorganization of the economy." On the other side of the ledger, Jamie Dimon of JPMorgan has called the spending "worth every dollar." Both quotes are on the record. They can't both be right; the listener decides which.
Now the part underneath it all — is there hidden inflation, a weak dollar? Here's what the data shows.
US M2 — the broad money supply — hit twenty-two point seven trillion dollars this spring, a record. For scale: from 2000 to 2019, M2 grew about six percent a year. In 2020 it grew nineteen percent, and another sixteen in 2021. That money didn't vanish. One read, from market commentators at 24/7 Wall Street: that wall of liquidity helps explain why risk assets keep climbing despite sticky inflation and high yields — and that, quote, "liquidity-driven rallies can become fragile when valuations outrun earnings growth."
The dollar side. The dollar index has been broadly weak through 2026, and strategists like those at ALPS are openly describing a "structurally weaker dollar" as a multi-year tailwind for assets priced outside it. When the unit of measurement shrinks, the price tags grow — that's mechanical, not magic. So part of "a trillion dollars" may be the trillion, and part may be the dollar.
One more fact, because it's the tell some analysts watch. Back in November, OpenAI's then-CFO Sarah Friar was asked how the company would finance its chip commitments. She mentioned banks, private equity, and a federal — quote — "backstop." Within twenty-four hours, three people walked it back: Friar herself, Sam Altman, and the White House AI czar David Sacks, who said flatly "there will be no federal bailout for AI." Reuters and others reported all three denials. Three denials, one day, same point — that's the sequence.
So the picture. Three record capital events in ten days. A financing structure that several major institutions describe as circular. A revenue gap that JPMorgan puts at twenty-six-fold. And a monetary backdrop — record M2, a softer dollar — that inflates every number you just heard. Whether that's a bubble, a weak dollar, hidden inflation, or a genuine bet on a technology that hasn't paid off yet — that's the debate, and the people in it disagree out loud.