Cerebras Beat Earnings, Stock Tanked 10%—Welcome to the IPO Divorce

Cerebras reported its first earnings as a public company last week and did what every growth stock is supposed to do: beat Wall Street projections. Revenue nearly doubled to $193.4 million from $99.5 million a year earlier. The stock fell 10% in extended trading anyway.

This is what happens when the market prices you at 90-100x revenue and then you remind everyone that margins exist.

The numbers looked fine on the surface. A 92% revenue increase in the first quarter post-IPO is objectively respectable for an AI chipmaker riding the current wave of infrastructure spending. Cerebras even guided to 88% core revenue growth in Q2, expecting to reach $914 million. For context, the company raised over $6 billion in its May offering—the largest U.S. tech IPO since Uber in 2019—and secured a deal worth over $20 billion with OpenAI. The narrative was intact. The thesis held.

Then the company announced that gross margins would compress to between 36% and 38% in the second quarter from 46.5% in the first quarter. That is not a rounding error. That is a 850 basis point drop. And that is the moment the stock began its 10% descent into the reality of running a hardware company at scale.

Cerebras blamed the margin squeeze on infrastructure. The company is temporarily renting back its own data-center capacity from a customer to meet near-term demand while it builds additional capacity. This is, in the language of CFO euphemism, a temporary measure that reflects aggressive guidance rather than structural weakness. It is also, in the language of equity investors who bought into an IPO priced at $185 only to watch it open at $350, a margin compression. The distinction matters less on a trading floor than it does in a earnings call transcript.

Morgan Stanley's Joseph Moore maintains an "Overweight" rating on the stock and lifted his price target to $273 from $250, arguing that the softer margin outlook reflects conservative guidance rather than weakening demand. This is the view of someone paid to believe in narratives. It may even be correct. But it does not explain why investors who bought at the opening price paid 1.9 times the IPO valuation for a company that immediately signaled margin compression. They did so because post-IPO enthusiasm follows a particular script: revenue growth validates the thesis, margin expansion validates the investment, and the stock does what stocks are supposed to do. Cerebras hit the first beat. It whiffed on the second.

The stock has now fallen 28% from its opening price, closing at $226.72 before the earnings report. This is the divorce settlement between IPO hype and quarterly reality. The hype priced in exponential margin expansion. Reality priced in temporary infrastructure rentals and the physics of scaling hardware production. One of these narratives was true. The market is behaving as if it now knows which one.

For a company with deeply negative operating margins, the margin expansion story was never optional. It was the entire thesis. Beat the revenue number, fine. Guide margins lower, and you have not beaten expectations at all. You have merely confirmed that the previous ones were wrong. And nothing spoils a growth narrative faster than the realization that someone miscalculated the denominator.