Cerebras Systems IPO 2026: Date, Price, Valuation, and Whether CBRS Is Worth Buying

Cerebras builds wafer-scale AI chips. A standard semiconductor chip is a small die cut from a silicon wafer. Cerebras's Wafer-Scale Engine (WSE-3) uses the entire wafer as a single chip, roughly 57 times larger than the largest competing GPU die, giving it vastly more on-chip memory, more processing cores, and higher interconnect bandwidth, all without the latency of communicating across a multi-chip GPU cluster. The flagship CS-3 system is designed for large-scale AI model training and inference and is sold as a turnkey compute environment with its own software stack. The total cost of ownership pitch, fewer systems, less power, less cooling infrastructure, is what attracted hyperscalers and enterprises. The most prominent win is a multi-year agreement with OpenAI for 750 megawatts of AI compute capacity, one of the largest AI infrastructure contracts ever disclosed.

At $135 per share, Cerebras is priced at approximately 51 to 53 times its 2025 trailing revenue of $510 million. By traditional metrics that is expensive. But Cerebras is profitable, $238 million in net income on $510 million in revenue is a 47% net margin that few IPO-stage technology companies can match. CoreWeave, the nearest comparable, went public in March 2026 at a $23 billion valuation on GPU cloud revenue and is not yet profitable at the same scale.

The 20x oversubscription at the original $115 to $125 range tells you institutional investors are willing to pay the premium. Whether the premium is earned depends on three things post-IPO: whether revenue growth accelerates beyond 76%, whether gross margins expand as the software and cloud services mix grows, and whether Cerebras adds a second major hyperscaler customer beyond OpenAI to reduce concentration risk.

Nvidia controls approximately 70 to 80% of the AI accelerator market and its CUDA software ecosystem creates switching costs that are measured in years of engineering investment, not hardware cost. Cerebras does not attempt to displace Nvidia across the board. The most defensible niche for Cerebras is capturing workloads where its wafer-scale architecture offers an inherent edge: massive model training jobs, serving models that demand extreme memory capacity, and operations where physical footprint is at a premium like sovereign AI, military use cases, or dedicated data centers. AMD's Instinct MI400 and Google's TPU v5 will also vie for these tasks. The single biggest threat to Cerebras's story is if Nvidia keeps shrinking the technical advantage gap generation by generation while its CUDA moat remains intact.

Cerebras Systems (CBRS) is projected to list on Nasdaq May 13 to 14, 2026. The share price has been revised up to the $125 to $135 range from an original $115 to $125, in response to over 20 times demand. At $135 per share, proceeds of around $3.78 billion could be generated at an all-in valuation of roughly $26 billion to $27 billion. Morgan Stanley, Citigroup, Barclays, and UBS are the syndicate lead managers.

As an IPO in the AI sector, Cerebras displays a distinctive set of strengths: revenue growth of 76% with real profitability to boot, with 47% net income margins ($238 million in net earnings and $510 million in top-line revenue). The OpenAI multi-year $750 MW agreement gives Cerebras revenue predictability along with tier-one industry acceptance. The biggest potential pitfalls include a highly concentrated customer portfolio (with OpenAI estimated to account for a very large portion of Cerebras revenue), the deeply-rooted nature of the Nvidia CUDA ecosystem, and a trailing revenue multiple of 51x (which will demand an equally multi-year performance from the company). The first structural challenge comes with a lock-up termination in mid-2026. The bullish story is that Cerebras will become the next large-scale AI infrastructure company. The bear view is that it remains a small player dependent on a single customer.

In 2025, Cerebras brought in revenues of $510 million (up 76% year-on-year) and earned a net profit of $238 million. Income comes through hardware sales (CS-3 AI supercomputer systems), recurring software services (including cloud-based AI services), and multi-year compute arrangements with hyperscalers and corporate enterprises. The OpenAI compute commitment for 750MW is the biggest public contract announced. Full disclosures about segmental revenues, customer mix, and cost components are contained in Cerebras's S-1 prospectus on EDGAR.

Cerebras Systems (CBRS) will list and trade on Nasdaq, beginning May 13 to 14, 2026. IPO units priced at the offering amount are being distributed to institutional investors and other qualified retail account holders via the underwriter syndicate. Retail participants will be able to buy Cerebras (CBRS) after the IPO has started on Nasdaq using any online stockbroker that has access to the exchange. Verify the real-time listing status at nasdaq.com/market-activity/ipos before submitting your buy order.

Cerebras is profitable, is growing at 76%, and it already has one of the biggest AI data center deals in history. The fact that the offering received 20-times oversubscription and its price band was lifted suggests this is a top-tier IPO for an AI company.

A 51 times trailing revenue valuation does demand proof of progress, and that is in the areas of revenue growth momentum, improved margins, and reducing reliance on OpenAI to broaden its client portfolio.

Observe the opening price of the shares on May 13 to 14 versus the $125 to $135 pricing window; look at the first quarterly earnings released since the shares began trading, and pay attention to any new hyperscaler deals it closes. This is the AI chip IPO of 2026.