53 minutes ago
2
Cerebras Systems, the AI chip company known for building processors the size of dinner plates, is raising its IPO price range to $150-$160 per share. That would value the company at roughly $32 billion, a staggering jump from its $4 billion private valuation in 2024.
The company plans to list on May 14, and the demand picture tells you everything you need to know: the offering is 20x oversubscribed. For context, that means investors want to buy twenty times more shares than Cerebras is actually selling.
From $4B to $32B in under two years
Cerebras had previously raised $720 million in private funding from investors including Alpha Wave Global and G42. Its last pre-IPO valuation sat at $4 billion in 2024, meaning the company’s implied value has grown eightfold before a single share trades on the public market.
AI chip IPO valuations have surged 150% year-to-date in 2026, and projected AI capital expenditures are expected to exceed $500 billion through 2028.
The company’s flagship product, the CS-3 wafer-scale chip, delivers 125 petaflops of AI compute. It can perform 125 quadrillion calculations per second, which is the kind of raw power that makes large language models and complex scientific simulations possible at scale.
Cerebras has deployed its technology in over 100 clusters globally. Its client list includes GlaxoSmithKline and the UAE government.
Taking a swing at Nvidia
In April 2026, Cerebras announced its CS-3 AI100 Inferencing System. The company claims it delivers inference speeds twice as fast as Nvidia H100 clusters at 40% lower cost.
Cerebras’ approach is fundamentally different from Nvidia’s. Rather than networking thousands of smaller GPUs together, Cerebras builds a single wafer-scale chip, essentially using an entire silicon wafer as one massive processor. The payoff is reduced communication bottlenecks between chips, which matters enormously for training and running large AI models.
What this means for investors
The $150-$160 price target puts Cerebras in rare company for a tech IPO. A $32 billion debut valuation would make it one of the largest US technology IPOs in recent memory.
Investors should pay close attention to revenue trajectory and gross margins once the company starts reporting as a public entity. Building wafer-scale chips is extraordinarily expensive, and the yields on such massive silicon are notoriously difficult to optimize.
With AI capital expenditures projected above $500 billion through 2028, hyperscalers like Microsoft, Google, and Amazon have every incentive to diversify their chip supply chains beyond a single vendor.
Cerebras has earned some investor confidence with real products and real customers, but an 8x valuation jump in under two years leaves very little margin for error.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) · 