Cerebras Systems Eyes $26.6B IPO Valuation
Cerebras Systems moves closer to its public debut. The AI chip company announced plans to offer 28 million shares priced between $115 and $125 apiece. At the top of that range, the IPO would generate $3.5 billion in proceeds and value the firm at $26.6 billion.
This pricing marks a quick gain for investors in the company's February $1 billion Series H round, which valued Cerebras at $23 billion. The listing also promises rewards for OpenAI and some of its leaders who hold investments.
A successful IPO at or above the high end would rank as 2026's biggest tech debut to date. Such performance might signal strong interest in upcoming large offerings from companies like SpaceX, and potentially OpenAI and Anthropic.
Cerebras' AI Chip Technology
Cerebras builds specialized hardware for AI workloads with its Wafer-Scale Engine 3 chip. The company claims this processor outperforms GPU alternatives in inference tasks while consuming less electricity. Inference refers to the computation required to handle user inputs in AI systems.
Cerebras positions its wafer-scale design as a direct rival to dominant GPU solutions in AI training and deployment.
Major Investors in Line for Gains
Several prominent venture firms hold significant positions in Cerebras. According to the firm's SEC filing, the top shareholders with over 5% each include Rick Gerson's Alpha Wave, Benchmark through partner Eric Vishria, Lior Susan's Eclipse, Fidelity, and Foundation Capital via partner Steve Vassallo.
The company lists additional backers such as 1789 Capital, Abu Dhabi Growth Fund, Abu Dhabi's G42, Alpha Wave Global, Altimeter, AMD, Atreides Management, Coatue, Moore Strategic Ventures, Tiger Global, Valor Equity Partners, and VY Capital.
Cerebras also highlights angel investors on its site. Among them are OpenAI CEO Sam Altman, OpenAI president Greg Brockman, former OpenAI chief scientist Ilya Sutskever who now runs his own AI venture, Quora CEO and OpenAI board member Adam D'Angelo, Sun Microsystems and Arista co-founder Andy Bechtolsheim, Intel CEO Lip-Bu Tan, and other industry figures.
OpenAI founder Sam Altman appears in the S-1 filing, though his stake falls below disclosure thresholds.
Stay ahead of the AI curve
The most important updates, news, and content — delivered weekly.
No spam. Unsubscribe anytime.
Close Ties to OpenAI
Cerebras maintains a deep connection with OpenAI beyond individual investments. Elon Musk cited this partnership in his legal action against OpenAI. Court documents from Musk's side note that OpenAI once explored buying Cerebras, and he claimed ignorance of OpenAI leaders' personal stakes.
No acquisition occurred, but OpenAI emerged as a key client. In December, OpenAI extended a $1 billion loan to Cerebras, backed by warrants for more than 33 million shares, as revealed in the S-1. This positions OpenAI to potentially become a major owner post-IPO.
The partnership includes a multi-year contract valued at over $10 billion.
Path to Public Markets
Cerebras targeted an IPO in 2024, but a U.S. government review of funding from Abu Dhabi cloud firm G42 derailed it. G42 remains a significant customer, the company states. That effort was abandoned.
In September of the following year, Cerebras secured $1.1 billion at an $8.1 billion post-money valuation, led by Fidelity and Atreides Management. Months later came the OpenAI deal with its loan and warrants. February brought the $1 billion Series H as the final large private raise.
Demand for the IPO appears robust. Bloomberg reports banks have fielded $10 billion in orders for the $3.5 billion share offering. This suggests Cerebras may price above the indicated range, boosting proceeds and shareholder value.
OpenAI and its team could profit further if the IPO succeeds, given their exposures.
Cerebras, founded in 2016, pioneered massive wafer-scale chips to accelerate AI computing, challenging Nvidia's GPU dominance in data centers. OpenAI, established in 2015, develops advanced models like GPT series and relies on vast compute resources for training and inference.