Broadcom Made $9.3 Billion. The Stock Fell 14%. SpaceX Lost $4.9 Billion. The Roadshow Began.

On Wednesday evening, Broadcom reported $9.31 billion in net income for a single quarter, up 88% year-over-year. The stock dropped roughly 14% on Thursday. On Thursday morning, SpaceX launched the roadshow for the largest IPO in history. The target: a $1.75 trillion valuation on $18.67 billion in annual revenue and a $4.94 billion net loss.

At 4:15 p.m. Eastern on Wednesday, Broadcom reported its fiscal second quarter. Revenue grew 48% year-over-year. Net income rose 88% to $9.31 billion. The company guided next quarter to $29.4 billion, nearly a billion dollars above the Street’s estimate. Jefferies raised its price target to $550.

By Thursday morning, the stock was down roughly 14%. The reason: CEO Hock Tan held the company’s $100 billion full-year AI revenue target. He did not raise it. The market had wanted $110 billion, maybe $120 billion. It got confirmation of $100 billion and sold.

What $100 billion means. Broadcom is on pace to generate $100 billion in AI chip revenue in a single fiscal year. That number, by itself, would have been inconceivable 18 months ago. It represents roughly 10% of global AI infrastructure spending projected for 2026, per Goldman Sachs. The company designs custom silicon for Google’s TPU chips under a contract running through 2031. It builds networking ASICs that connect the clusters where large language models train. It is the plumbing company for the AI buildout. And $100 billion was not enough.

The stock fell because the expectations embedded in the share price assumed the target would be raised. When it was not, the gap between the narrative and the number became visible. The narrative said AI spending is accelerating without limit. The number said it is accelerating at the rate Broadcom already told you it would. That distinction cost shareholders roughly 14% of the company’s market capitalization in a single session.

What $1.75 trillion means. Four hours after Broadcom’s stock began falling, SpaceX launched its investor roadshow at a fixed price of $135 per share. The company plans to raise $75 billion at a valuation of $1.75 trillion. That would make it the largest IPO in history, more than double Saudi Aramco’s 2019 record. SpaceX reported $18.67 billion in 2025 revenue and a net loss of $4.94 billion, per its S-1/A filed June 1. At $1.75 trillion, the stock would trade at roughly 94 times trailing revenue. Morningstar’s June 1 research note valued SpaceX at $780 billion, 55% below the IPO target.

The company is not yet profitable. Its core Starlink business generated $11.4 billion in revenue and is the only segment in the black. The launch business grew 8%, and most Falcon 9 missions served internal Starlink deployments rather than paying customers. The long-term thesis rests on solar-powered AI data centers in orbit, a technology that does not yet exist. SpaceX calls it a “$28.5 trillion addressable market,” per Reuters.

The 24-hour gap. These two events happened in the same news cycle. A company that earned $9.31 billion in a quarter was punished for not promising more. A company that lost $4.94 billion in a year was rewarded with the largest public offering in market history. Broadcom is being priced on what it delivers. SpaceX is being priced on what it describes. The market is running two valuation systems simultaneously, and they produce opposite conclusions about what earnings are worth.

The base case here is not that SpaceX is overvalued or that Broadcom was unfairly sold. It is that the market has entered a phase where the bar for established AI companies keeps rising. The bar for narrative-driven companies stays on the ground. Broadcom needed to beat and raise to hold its price. SpaceX needs only to describe a future to command $1.75 trillion. The question for the reader holding an AI-heavy portfolio is which valuation system applies to the rest of their holdings.

My read: the Broadcom sell-off and the SpaceX roadshow are two sides of the same coin. The market is telling you that $100 billion in real AI revenue is not enough to sustain a stock price. It is also telling you that $18.67 billion in revenue and a net loss is enough to justify the biggest IPO ever. One of those frameworks is wrong. Friday’s payrolls report will not resolve this. But it will tell us whether the economy underneath both bets is still hiring. And that, eventually, is the number that decides which framework survives.