Cerebras IPO: $5.55B Raise, 68% Debut, $95B Market Cap

The trading day

Cerebras priced its offering at $185 a share, above a range that had already been raised once, and sold 30 million shares to raise $5.55 billion. The stock opened around $350 and touched an intraday high near $386 before settling to close at $311.07, up 68% from the IPO price, for a market capitalization near $95 billion. The listing traded on Nasdaq under the ticker CBRS, led by Morgan Stanley, Citigroup, Barclays and UBS. [1]

CNBC's reporting on the day called it the largest U.S. tech listing since Uber's 2019 debut, and placed it against a thin backdrop: 31 tech IPOs priced in 2025, against 121 four years earlier. A single-product chip company briefly outdrew that entire market's recent pace for a day. [1]

Why this listing almost did not happen in 2026

Cerebras filed to go public once before, in 2024, and shelved those plans. TechCrunch's reporting on the 2026 listing attributes the delay to scrutiny from the Committee on Foreign Investment in the United States over a large investment from G42, the Abu Dhabi AI holding company, and to concern about how much of Cerebras's revenue depended on that one relationship. Getting from a shelved 2024 filing to a priced 2026 offering meant resolving both. [2]

The financials in the same reporting show the resolution was mostly about growth diluting concentration rather than G42 disappearing as a factor. Cerebras reported 2025 revenue of $510 million, up 76% year over year, and net income of $237.8 million, a reversal from a 2024 net loss of roughly $500 million. Customers named in that coverage include OpenAI, G42, the Mohamed bin Zayed University of Artificial Intelligence, and Amazon Web Services. [2]

Atlas interpretation: A company that swings from a half-billion-dollar loss to a quarter-billion-dollar profit in one year, while growing revenue three-quarters over, has a much easier time explaining away a concentrated early customer than one still losing money. The CFIUS review was about who Cerebras depended on. The two years in between mostly changed the answer by adding OpenAI and AWS next to G42, not by removing G42. [2]

What the pop signals, and what it does not

Atlas interpretation: A 68% first-day close, after an even bigger intraday spike, is a pricing failure as much as a demand signal: Cerebras and its underwriters left billions of dollars that could have gone to the company on the table by pricing too low. That is a normal feature of hot IPOs, not evidence on its own that a $95 billion valuation for a company with $510 million in revenue is sound. Public trading, not the first session, is what tests that. [1][2]

Atlas interpretation: The more durable fact in the reporting is the drought it broke rather than the multiple it set. A market that priced 31 tech companies in all of 2025 absorbing one AI chipmaker's offering with this much demand says more about pent-up appetite for AI-adjacent public equity than it does about Cerebras's wafer-scale chips specifically being worth that price. Whether more AI infrastructure companies follow Cerebras onto Nasdaq in 2026 is the test of that reading. [1]