SpaceX IPO'd at $1.77 trillion and closed its first day at $2.1 trillion, making it the largest public offering in history. The company lost nearly $5 billion last year, has never turned a profit on rocket launches despite operating since 2002, and is spending heavily on AI infrastructure it does not yet know how to fully deploy. The valuation is not justified by the books. It is justified by something else.
That something else is the structure of how moonshot capital works. Investor willingness to absorb years of losses on improbable bets, satellite internet, space data centers, frontier AI, is what turns those bets into real infrastructure. The waste is real. So are the results. The same logic applies to the broader AI buildout: some of today's record capex will be zeroed out, OpenAI is already signaling price cuts to fight Anthropic for users, data center deals are collapsing, and SpaceX itself is selling compute to Google rather than using it internally. The overexuberance is funding the progress.
The part worth reading closely is who absorbs the downside when these bets go wrong. Retail investors took somewhere in the low-20% range of the SpaceX IPO, versus a typical 5 to 10%. Everyday buyers are increasingly at the table for both the upside and the write-downs. The original piece works through the tension between necessary irrationality and who actually pays for it when the math catches up.
[READ ORIGINAL →]