AI chip specialist Cerebras Systems priced its IPO at $185 per share on May 13–14, 2026, above its raised range of $150–160, after 20x oversubscription during roadshow. The company raised $5.55 billion and attained a fully diluted market cap of approximately $56.4 billion. Shares opened at $350 on day one—nearly double the IPO price—before closing at $311.07, then fell to $180 by June 4 as day-one euphoria correction took hold.
Cerebras designs specialized AI chips around its Wafer Scale Engine (WSE-3), featuring 4 trillion transistors on a single silicon wafer the size of a dinner plate. The pitch is faster and cheaper inference performance than Nvidia GPUs for certain AI workloads, positioning the company as a speed-and-cost challenger in the inference market—not training. A March AWS partnership and OpenAI's $20 billion-plus compute deal solved the prior customer-concentration problem; OpenAI has handed Cerebras warrants worth up to 10% of the company (around $5 billion at IPO midpoint), roughly half the gross profit OpenAI will make on the deal.
For practitioners, Cerebras' IPO success reflects investor appetite for Nvidia alternatives in inference-at-scale, but the stock's 44% pullback from day-one high warns that valuations are frothy. The real question: can Cerebras maintain reference customers and build enterprise deployment volume faster than Nvidia can pivot into inference optimization? Product delays or yield issues on wafer-scale chips could evaporate the hype. Teams evaluating inference infrastructure should benchmark Cerebras against AMD, Groq, and native Nvidia inference options, not assume day-one momentum reflects fundamental advantage.