Cerebras Surges 68% in Blockbuster Nasdaq Debut, Valuing AI Chipmaker at $95 Billion

Cerebras Systems, maker of the wafer-scale AI chip, priced its IPO at $185 and surged nearly 70% on its first day of Nasdaq trading, raising $5.55 billion and briefly pushing…

Cerebras Surges 68% in Blockbuster Nasdaq Debut, Valuing AI Chipmaker at $95 Billion

Overview

Cerebras Systems made its Nasdaq debut on May 14, 2026, in what immediately became the most dramatic and closely watched technology IPO of the year. The company priced its shares at $185 the night before, raising $5.55 billion from 30 million Class A shares — the largest US tech IPO since Uber’s market debut in 2019. When the opening bell rang, shares began trading near $350, nearly double the IPO price, and the stock closed the first session at $311, representing a 68% gain and giving Cerebras a market capitalisation of approximately $95 billion. The debut minted two new billionaires: CEO Andrew Feldman and CTO Sean Lie, who own stakes valued at $3.2 billion and $1.7 billion respectively at closing prices.

What Cerebras Actually Builds

Cerebras is an AI chip company built on a fundamentally different architecture from every competitor in the market. Where standard GPUs and AI accelerators are cut from silicon wafers into individual chips and then reconnected through high-speed interconnects, Cerebras treats an entire wafer as a single processor. Its Wafer Scale Engine 3 (WSE-3) is approximately 58 times larger than a leading GPU die, containing around four trillion transistors on a dinner-plate-sized chip. The design eliminates the inter-chip communication bottleneck and packs enormously more SRAM on a single die — delivering up to 21 petabytes per second of memory bandwidth, nearly 1,000 times faster than Nvidia’s Rubin GPUs.

That architecture makes Cerebras’s systems particularly well-suited for AI inference workloads — the phase where trained models respond to live user queries — and the company consistently ranks among the fastest inference providers globally, generating over 2,200 tokens per second on large models in some benchmarks.

The OpenAI Deal That Built the Bull Case

The central pillar of investor enthusiasm is Cerebras’s $20 billion multi-year compute deal with OpenAI, announced in early 2026, covering 750 megawatts of inference capacity with options to expand to two gigawatts by 2030. Combined with a March 2026 agreement in which AWS will deploy Cerebras CS-3 systems inside Amazon data centres and make them available through Amazon Bedrock, the company entered its IPO with two of the most credible enterprise AI customers in the world providing revenue visibility.

The offering was reportedly more than 20 times oversubscribed, and the initial pricing range of $115 to $125 was raised to $150 to $160, then to $185 on the final evening. Benchmark Capital, an early Series A investor, saw its stake rise to $5.5 billion at closing.

Risks and What Comes Next

Despite the euphoria, risks are substantial. Customer concentration is severe: approximately 86% of 2025 revenue came from two UAE-linked entities. The company’s core operating business ran at a loss in 2025, and the stock at closing trades at over 130 times annual sales. Cerebras’ WSE-3 architecture is also approaching a refresh cycle, and Nvidia’s broader ecosystem advantages remain formidable.

Still, the IPO has opened the door for a potential wave of AI infrastructure public listings, with OpenAI, Anthropic, and SpaceX all cited as candidates for major offerings later in 2026.

Share:
Subscribe
Notify of
0 Comments
Inline Feedbacks
View all comments

Discover More

Introduction to iOS Settings: Configuring Your Device

Learn how to configure and manage your iPhone or iPad with this comprehensive guide to…

Datatruck Raises $12M to Build AI Operating System for Trucking

Logistics startup Datatruck raises $12 million Series A to modernize trucking operations with predictive routing…

Building a Personal Brand as a Data Scientist

Building a Personal Brand as a Data Scientist

Learn how to build a strong personal brand as a data scientist. Discover strategies for…

AI-Powered Human Trafficking Detection Tools Enter Operational Law Enforcement Deployment

AI-Powered Human Trafficking Detection Tools Enter Operational Law Enforcement Deployment

Advanced data science models trained to identify trafficking networks through pattern recognition across communications, financial,…

Implementing Logistic Regression with Scikit-learn

Implementing Logistic Regression with Scikit-learn

Learn to implement logistic regression with scikit-learn step by step. Covers solvers, regularization, multi-class, hyperparameter…

Introduction to the Linux Terminal: Why You Should Learn It

Introduction to the Linux Terminal: Why You Should Learn It

Discover what the Linux terminal is, why it’s more powerful than graphical interfaces, and how…

Click For More
0
Would love your thoughts, please comment.x
()
x