Cerebras Systems prices its initial public offering on 13 May on the Nasdaq Global Select Market under the ticker CBRS, in what Dealogic confirms is the largest IPO globally so far in 2026. The company raised the price range to $150–$160 per share from an initial $115–$125, and increased marketed shares from 28 million to 30 million — putting the total raise at approximately $4.8 billion and the fully diluted valuation at up to $48.8 billion.
The demand has been extraordinary. Orders exceeded available shares by more than 20 times, reflecting investor conviction that Cerebras has built a genuinely differentiated product in a market dominated by Nvidia. Morgan Stanley, Citigroup, Barclays, and UBS are underwriting the offering.
Cerebras reported $290.3 million in 2025 revenue — a 76 per cent year-over-year increase — with net income of $87.9 million, a dramatic turnaround from a $485 million loss the prior year. The company's core product is the WSE-3, a wafer-scale chip featuring 900,000 cores and a 44-gigabyte pool of SRAM optimised for AI inference workloads. Unlike Nvidia's GPU-based approach, Cerebras builds chips at the scale of an entire silicon wafer, eliminating the interconnect bottlenecks that limit traditional chip designs.
The customer base has diversified significantly. Early revenue was concentrated with UAE-based G42, but Cerebras has since added OpenAI and Amazon Web Services as clients — two of the most demanding AI infrastructure buyers in the world. The company also holds a 750-megawatt compute deal with OpenAI, one of the largest AI data centre agreements ever signed.
For context engineers, the Cerebras IPO signals that the AI hardware market is moving beyond Nvidia monoculture. The WSE-3's architecture is purpose-built for the inference workloads that power production AI applications — the exact workloads that context engineers rely on daily. If Cerebras can deliver on its inference performance claims at scale, it could meaningfully reduce the cost and latency of frontier model deployments, benefiting every developer who builds on top of API-served models.
