On June 11, 2026, SpaceX sold 555.6 million shares at $135 and raised $75 billion, the largest IPO in American history. The debut valued the company around $1.77 trillion. Retail investors scrambled for allocation, and the offering was structured to give them an unusually large piece. CNBC found buyers who called the valuation "stupid" in the same breath they used to describe buying it. Elon Musk sold zero shares.
There's an old piece of trading folk wisdom that record IPOs ring the bell at the top. Like most folk wisdom, it's directionally right and operationally wrong, and the difference between those two things is where accounts go to die.
So let's look at the actual record.
The bell rings near the top, not at it
The two examples everyone cites are AT&T Wireless and Visa, and both cut against the "sell that day" version of the story.
AT&T Wireless priced in April 2000 and raised $10.6 billion, the largest US IPO at the time. The NASDAQ had already peaked on March 10, six weeks earlier. The deal priced into the first bounce, the one that convinced everyone the dip was for buying. The index lost roughly 70% over the next two years.
Visa priced in March 2008 and raised $17.9 billion, a new record. The S&P had topped five months earlier, in October 2007. Visa priced two days after Bear Stearns collapsed. The market then rallied for two months into May 2008, punishing anyone who treated the deal as a sell signal, and then lost half its value.
Here is the fuller sample:
| Mega IPO | Date | Raised | What followed |
|---|---|---|---|
| AT&T Wireless | Apr 2000 | $10.6B | NASDAQ had topped 6 weeks earlier; minus 70% over 2 years |
| Blackstone | Jun 2007 | $4.1B | S&P top 4 months later |
| Visa | Mar 2008 | $17.9B | 2-month bear rally, then minus 50% |
| Glencore | May 2011 | $10B | Commodity supercycle top that week |
| Alibaba | Sep 2014 | $25B | Bull ran 5 more years under expanding QE3 |
| Saudi Aramco | Dec 2019 | $29.4B | COVID top 2 months later |
| Coinbase | Apr 2021 | Direct listing | Bitcoin topped the same day |
| Rivian | Nov 2021 | $11.9B | Speculative top that month; S&P top in Jan 2022 |
Seven of eight landed within a few months of a major top in their asset class. That's a remarkable hit rate for any single indicator. But notice two things the table actually says, because they're not what the folk wisdom says.
First, the window is months, not days. Sometimes the deal prices before the top (Blackstone, Aramco, Rivian) and sometimes after it (AT&T Wireless, Visa). What the record deal marks is not the price top. It marks the moment the people with the best information about demand, the issuers and their bankers, decided that public appetite would never be better. An IPO this size is a liquidity harvest. You harvest at peak ripeness, and peak ripeness is a season, not a day.
Second, look at the one clean miss. Alibaba raised $25 billion in September 2014 and the bull market ran five more years. What was different? The Federal Reserve was still running QE3 and global liquidity was expanding. Every hit on that table shares the opposite condition: the deal priced into tightening liquidity or within months of it. The bell only tolls when the water level is about to drop.
Where we are now
So the honest version of the indicator is conditional: a record IPO marks peak harvestable liquidity, and it precedes a major top when the liquidity regime is turning restrictive. Two questions follow. How big was the harvest, and which way is the water moving?
The harvest: $75 billion is about 2.8 times Visa's record in real terms. Tilted deliberately toward retail allocation. The single largest holder sold nothing into it. And it is explicitly not the end of the supply calendar: Anthropic and OpenAI are expected to come public behind it. In 1999 and 2000, the top didn't arrive with the first record deal. It arrived when the issuance calendar saturated demand, when the third and fourth mega-deals priced into an audience that had already spent its enthusiasm. One bell is a warning. Three bells into fading breadth is the event.
The water is where the framing has to get sharper than "CPI at 4.2%, FOMC next week." Two structural pieces are doing the actual work right now.
First, the rubberband. Rates went from 18% under Volcker in 1981 to roughly zero by 2020. Forty years of a single direction. That arc has turned, and we are five years into the reversal. Governments will try to suppress every move higher, they have to, given how much debt service is already a budget line, but the suppression behaves like a rubberband. Pull harder back, the snap forward is harder. SpaceX is harvesting into a regime where the structural force on the cost of capital is pulling against the suppression. The Alibaba era priced into expanding QE3 liquidity. This era prices into a Fed that cannot run that playbook without consequence.
Second, the compression. The crash that took 30 months to play out in 2000 took 23 trading days in 2020. The 2024 yen carry unwind played out in two weeks. The money-printer reflex is codified now, fiscal-monetary cooperation at speeds and sizes that did not exist pre-COVID. Which means the bell that took six weeks to mark the 2000 top can mark this one in days. The window I am describing as "months" is what 1999 would have called "days." Anyone waiting for a comfortable confirmation will get there after the move is over.
I don't know what the Fed will do at the June 16-17 meeting. I do know which column of the IPO table this configuration resembles, and it is not the Alibaba column.
What I'm watching instead of predicting
I publish research, not prophecy, so here is the checklist rather than the call:
The $135 line. AT&T Wireless broke its issue price within weeks in 2000, and that break was the real demand-exhaustion signal. If SPCX trades below issue this summer, the harvest-complete reading gets much stronger. If it holds and builds, the audience still has money and the season runs longer.
The next bells. How the Anthropic and OpenAI deals price, and what market breadth looks like when they do. Supply meeting strong demand is a bull market. Supply meeting tired demand is distribution.
The labor tape. Liquidity tops become price tops when the real economy confirms. Weekly jobless claims through late summer will tell that story faster than quarterly data.
What I'm not doing
Not shorting the euphoria. That's the operational lesson the table teaches: even when the bell was real, it paid the patient and carried out the eager. Anyone who sold Visa's IPO as a top signal in March 2008 sat through a two-month rally against them before being right. Being early on a regime turn feels identical to being wrong for exactly as long as your risk tolerance lasts, and usually one week longer.
The discipline that survives this part of the cycle is unglamorous: stay directional where the evidence is, size so that being months early is survivable, and let the checklist confirm instead of letting the headline excite. In a compressed-cycle regime, the temptation is to lean harder into the call because the move when it comes will be fast. That instinct is the inverse of what the table teaches. Fast moves both ways. Size for the rubberband direction; pre-position for the snap.
The biggest IPO in history is not a sell order. It's the sound of the smartest sellers in the world deciding this is the best audience they will ever get. I take that seriously without trading it impulsively.
There's a name for the broader structure this fits into. The engineered kind of euphoria that hands retail the bag at the exact moment insiders won't buy their own valuation. The two pieces above, a forty-year rubberband finally pulling forward and cycles that complete in weeks instead of years, are the two halves of the mechanism. That structure gets its own post.