Cerebras Systems filed, on 4th May, updated IPO terms with the SEC, targeting up to $3.5 billion in a Nasdaq listing at $115 to $125 per share. At the high end of that range, the Sunnyvale-based AI chipmaker would be valued at $26.6 billion when it prices under the ticker CBRS on May 13. It would be the largest tech IPO of 2026 so far, and the most significant public market test of whether investors are willing to fund a serious Nvidia alternative.
The Cerebras Systems IPO has been a long time coming. The company first attempted to go public in 2024 before withdrawing its paperwork as it pivoted away from selling hardware directly toward running its own cloud infrastructure. That restructuring turned out to be the right call. It returned to file an S-1 in April 2026 with a fundamentally different financial profile: $510 million in Q4 2025 revenue, up 76% year over year, and $87.9 million in net income for the same period.
What the Cerebras Systems IPO Is Actually About
Cerebras is not a GPU company. It builds wafer-scale chips, processors that occupy an entire 300-millimeter silicon wafer rather than the small individual dies that Nvidia stacks and connects in GPU clusters. The flagship WSE-3 is physically 57 times larger than Nvidia’s H100 and carries 2,625 times more memory bandwidth than Nvidia’s B200 package. The architecture is designed to eliminate the latency overhead that comes from coordinating thousands of individual GPUs wired together. For workloads where memory bandwidth is the binding constraint, Cerebras makes a credible performance argument.
Also Read: GalaxEye’s Mission Drishti: World’s First OptoSAR Satellite
That argument has attracted serious validators. The Cerebras OpenAI compute deal, announced in January 2026, commits OpenAI to purchasing 750 megawatts of Cerebras computing capacity through 2028 in a contract valued at over $20 billion. OpenAI also lent the company $1 billion, secured by warrants that could convert into a minority equity stake. Amazon Web Services signed a binding term sheet earlier this year to become the first major hyperscaler to deploy Cerebras chips inside its own data centers. AMD participated in the February 2026 Series H round, which valued the company at $23 billion.
These are not casual endorsements. They represent the three most consequential validators in AI infrastructure: the leading AI lab, the leading cloud provider, and a major semiconductor company. For the AI chip IPO 2026 market, the demand signal is unusually clear. Banks underwriting the deal have already received indications of interest exceeding $10 billion on a $3.5 billion offering, nearly three times oversubscribed before a single share has been priced.
The Risk the Prospectus Makes Clear
The S-1 disclosure that deserves the most attention is customer concentration. G42, an Abu Dhabi-based cloud provider, accounted for 87% of Cerebras revenue in the first half of 2024. The company’s entire bull case rests on successfully completing a customer transition from G42 to OpenAI and cloud hyperscalers. That transition is underway. It is not finished.
The financial entanglement with OpenAI also warrants scrutiny. OpenAI is simultaneously Cerebras’s largest customer, its lender, and a potential future shareholder through warrant conversion. That is a concentration of dependency that public market investors have not fully priced into the headlines. If OpenAI’s own strategic priorities shift, or if its in-house chip development program accelerates, the downstream effect on Cerebras revenue would be material.
CEO Andrew Feldman is not selling any shares in the offering. He will retain 10.3 million shares post-IPO, valued at up to $1.28 billion at the high end of the range. That is a meaningful signal of conviction. It is also worth noting that the wafer-scale chip Nvidia alternative thesis remains unproven at hyperscale deployment, outside of the OpenAI relationship.
What the Cerebras $26 Billion Valuation Implies
The Cerebras CBRS Nasdaq listing is pricing in a future that is directionally clear but not yet operational at scale. At $26.6 billion, investors are paying roughly 52 times trailing revenue on a company whose largest customer also holds warrants and sits on its cap table. By comparison, Nvidia trades at roughly 30 times forward revenue, with a dominant software ecosystem and broad customer diversification.
The gap is not irrational given Cerebras’s growth rate. But it leaves almost no margin for execution risk on the OpenAI transition. If the roadshow closes successfully on May 13, the Cerebras Systems IPO will confirm that public markets are prepared to fund the next layer of AI infrastructure investment, not just the applications built on top of it. If it stumbles at pricing, the implications for the broader AI chip IPO 2026 pipeline, including potential listings from other infrastructure players, will be significant.
The wafer-scale bet is technically sound. The business model transition is in progress. Both of those things can be true at the same time, and neither cancels the other out.
