Cerebras Systems set the price range for its Nasdaq debut Monday, filing to sell 28 million Class A shares at $115 to $125 apiece, according to an updated SEC prospectus. The offering would raise as much as $3.5B before a 4.2 million-share underwriter option. At the top of the range, the Sunnyvale, California, AI chipmaker would be valued at about $26.6B.

Key Takeaways

AI-generated summary, reviewed by an editor. More on our AI guidelines.

Banks start the roadshow

Morgan Stanley, Citigroup, Barclays and UBS Investment Bank are leading the sale, with Mizuho and TD Cowen acting as bookrunners. Cerebras has applied to list on the Nasdaq Global Select Market under the ticker CBRS.

The proposed listing follows a February Series H that brought in $1B at a post-money valuation of about $23B. Tiger Global led that round, and Cerebras named Benchmark, Fidelity Management & Research Company, Atreides Management, Alpha Wave Global, Altimeter, AMD, Coatue and 1789 Capital among the participants.

The IPO range also comes after Cerebras withdrew earlier paperwork in October 2025. CNBC reported at the time that the company wanted to update financials and strategy after shifting away from direct hardware sales toward cloud access built on its own chips.

OpenAI terms shape the pitch

According to the May 2026 amendment to its registration statement, Cerebras booked revenue of $510 million for 2025, up from $290.3 million the prior year. GAAP net income for the period came to $237.8 million, though on a non-GAAP basis the company still posted a loss of $75.7 million after adjustments. As of Dec. 31, the prospectus reported $24.6 billion in remaining performance obligations, of which Cerebras told the SEC it expects to recognize roughly 15% across fiscal 2026 and 2027.

OpenAI supplies a major forward commitment in the prospectus. Cerebras disclosed that the ChatGPT maker agreed to take up to 750 megawatts of computing power through 2028, with another 1.25 gigawatts available through 2030. CNBC wrote in April that the expanded OpenAI relationship includes a $1B loan from OpenAI to help Cerebras build data center infrastructure.

The OpenAI commitment gives Cerebras a customer story for the roadshow, while the revenue table keeps concentration risk in view. CNBC reported that G42 supplied 24% of 2025 revenue, while Mohamed bin Zayed University of Artificial Intelligence supplied 62%.

Wafer-scale chips remain the difference

Cerebras sells a different architecture from Nvidia's rack-scale GPU systems. Its Wafer-Scale Engine 3 is built from almost an entire silicon wafer and is marketed for low-latency inference as well as training. The company says the chip is 58 times larger than a leading GPU chip and can deliver inference up to 15 times faster than leading GPU-based systems on open-source models.

The offering arrives as cloud providers and model labs keep adding specialized silicon options. Google recently split its TPU 8 line into separate training and inference chips to attack serving costs, while Nebius bought Eigen AI to squeeze more tokens from each Nvidia chip. Cerebras is asking investors to price a public-market case built around wafer-scale inference rather than rented Nvidia capacity.

Roadshow moves toward pricing

Andrew Feldman, Cerebras' co-founder and chief executive, is not selling shares in the offering, according to the prospectus. His retained stake would be worth as much as $1.28B at the top of the range.

The underwriters' option could add $525M in proceeds if sold at $125 a share. Pricing is the next scheduled marker, when investors will decide whether Cerebras' OpenAI-backed order book offsets the customer concentration and infrastructure demands disclosed in its filings.

Frequently Asked Questions

How much does Cerebras plan to raise in its IPO?

Cerebras plans to sell 28 million Class A shares at $115 to $125 each. At the top of the range, the offering would raise about $3.5 billion before any underwriter option.

What ticker will Cerebras use on Nasdaq?

Cerebras has applied to list on the Nasdaq Global Select Market under the ticker symbol CBRS, according to its updated SEC prospectus.

Why does OpenAI matter to the Cerebras IPO?

OpenAI agreed to take up to 750 megawatts of Cerebras compute through 2028, with another 1.25 gigawatts available through 2030. CNBC also reported a $1 billion OpenAI loan tied to data center infrastructure.

What is Cerebras’ main technical difference from Nvidia?

Cerebras builds wafer-scale processors rather than conventional GPU clusters. The company says its WSE-3 chip is 58 times larger than a leading GPU chip and targets low-latency inference as well as training.

What risk does the filing highlight?

The filing and CNBC coverage point to customer concentration. In 2025, CNBC reported that G42 supplied 24% of revenue and Mohamed bin Zayed University of Artificial Intelligence supplied 62%.

AI-generated summary, reviewed by an editor. More on our AI guidelines.

Nvidia still wins per chip. Google just changed what counts.
Google Cloud on Wednesday unveiled two new TPUs at Cloud Next 2026, splitting its eighth-generation design into a training chip and an inference chip for the first time in the program's decade-long hi
Meta Turns to Amazon Graviton Chips as AI Agents Shift Compute Math
Meta has signed a multiyear deal to run AI workloads on Amazon's Graviton processors, giving AWS one of its largest outside validations for homegrown server chips. The deployment begins with tens of m
Google splits TPU 8 into training and inference chips to chase Nvidia
Google unveiled its eighth-generation tensor processing unit in two distinct variants Wednesday at Cloud Next 2026, splitting training and inference work onto separate silicon for the first time in th
AI News

San Francisco

Editor-in-Chief and founder of Implicator.ai. Former ARD correspondent and senior broadcast journalist with 10+ years covering tech. Writes daily briefings on policy and market developments. Based in San Francisco. E-mail: [email protected]