SpaceX S-1 — One Profitable Segment, a $1.77T Price Tag

The setup

This is not a quarterly print — it's the first time anyone outside the cap table has seen SpaceX's audited financials. The company filed its S-1 with the SEC on May 20, 2026, launched its roadshow June 4, and is set to price after the close June 11 to trade June 12 on Nasdaq under SPCX. At a fixed $135 and 555.6M Class A shares, it's a ~$75B raise at roughly a $1.77 trillion valuation — the largest IPO in history by a wide margin, with an underwriters' over-allotment of a further 83.33M shares (~$11.2B) on top. We treat the S-1 the way we'd read any first look at a company that has been a black box for two decades.

The headline tension is simple: Starlink has crossed over into being a genuinely profitable, fast-compounding broadband business — and the consolidated entity still lost nearly $5B, because SpaceX folded in xAI (acquired February 2026) and inherited its losses. You are underwriting two companies stapled together: a cash machine and a cash furnace.

The numbers

Total revenue of $18.7B in 2025, up 33% from $14.1B in 2024. Starlink did $11.4B of that, growing 49.8% — still fast, but decelerating hard from 96.4% in 2024, which is the number bulls should sit with. Starlink threw off $4.4B of GAAP operating income at a 39% margin (and a 63% non-GAAP segment-adjusted EBITDA margin), so the broadband unit is no longer a story, it's a business. Consolidated adjusted EBITDA was $6.58B.

The bridge from $6.58B of adjusted EBITDA to a $4.94B GAAP net loss runs almost entirely through the AI segment, which posted a $6.35B operating loss for the year. Strip xAI out and SpaceX-classic — launch plus Starlink — is a profitable, cash-generative enterprise. Bolt xAI on and the consolidated P&L is deeply red. The whole valuation debate is really a debate about how much of that AI loss you believe converts into a defensible franchise versus a permanent subsidy.