Cerebras Systems Inc. (NASDAQ:CBRS) begins trading today on the Nasdaq at $185 per share, carrying a $56.4 billion fully diluted valuation into its first morning as a public company. That number is not the story. Rather, the story is what investors are paying for versus what the business actually delivers today, and the gap between those two things is wider than the first-day excitement will suggest.
Institutional Demand Signals Sentiment, Not Value
A 20-times oversubscribed book sounds like validation. In reality, it is not. Specifically, it tells you that institutions wanted more shares than were available at a fixed price, which is a function of supply scarcity and AI sector momentum. The most oversubscribed IPOs in history have both soared and collapsed on day one. Put simply, oversubscription measures demand for an allocation, not conviction in a valuation.
Retail investors entering CBRS at the open today are not getting the institutional price. Instead, they are absorbing the premium that institutional crowding created. The question is not whether AI infrastructure is real. It is whether this valuation, on this revenue base, leaves room to generate a return.
The Revenue Base Does Not Match the Price Tag
At $185, CBRS trades at roughly 110 times its 2025 revenue of $510 million. Using the company’s own $24.6 billion backlog, with management projecting 15% recognition across 2026 and 2027, annualized forward revenue approaches $1.85 billion. That puts the forward price-to-sales multiple at approximately 30 times. By comparison, NVIDIA Corporation (NASDAQ:NVDA) trades at roughly 20 times forward revenue, with proven profitability, a dominant software ecosystem, and a customer base spanning every major hyperscaler.
Moreover, the GAAP net income of $237.8 million that appears in headlines came almost entirely from a one-time non-cash gain of $363.3 million on a forward contract liability extinguishment tied to G42, per the Cerebras S-1/A filed May 4, 2026. Strip that out and …
This post was originally published here
