Cerebras is bringing one of the year’s biggest artificial intelligence chip listings to market. The IPO will test whether investors still want to pay premium prices for companies challenging Nvidia’s grip on AI infrastructure.

The Sunnyvale-based chipmaker is seeking to raise as much as $3.5 billion in a U.S. initial public offering, according to reports. Cerebras plans to sell 28 million shares at $115 to $125 each. The company expects to list on Nasdaq under the ticker CBRS.

The IPO comes at a moment when AI infrastructure remains one of the strongest investment themes in global markets. Demand for chips, data centers and inference capacity has turned semiconductors into a central market story, not only a technology story. Cerebras is trying to position itself as one of the few visible alternatives to Nvidia in high-performance AI computing.

The company’s pitch is built around its wafer-scale engine chips, which are designed to accelerate training and inference for large AI models. Cerebras argues that its architecture can reduce memory and data-movement bottlenecks that affect conventional GPU systems. That message matters because AI workloads are becoming larger, more expensive and more dependent on specialized hardware.

The financial growth is sharp. Cerebras reported revenue of $510 million for 2025, up from $290.3 million a year earlier. The company also moved into profit on a per-share basis after posting a loss the previous year.

But the market question is not only whether Cerebras is growing. It is whether public investors will accept a valuation that prices in years of AI demand before the company has proved it can scale like a mature infrastructure provider.

That is where the IPO becomes a broader test for markets. Private AI valuations have moved faster than public-market discipline. A successful Cerebras listing would show that investors still want exposure to AI hardware beyond Nvidia, even at aggressive multiples. A weak debut would suggest that the AI trade is becoming more selective.

There are also concentration risks. Cerebras relies heavily on large customers and major AI infrastructure contracts. That can support rapid revenue growth, but it also makes the company more exposed if spending patterns change, contracts are delayed or customers shift to competing systems.

For the IPO market, the deal matters because it arrives at the intersection of three powerful themes: AI spending, semiconductor scarcity and investor demand for new public growth names. If Cerebras prices well, it could open the door for more AI infrastructure companies to test public markets.

For Nvidia, the threat is not immediate displacement. Nvidia still owns the dominant ecosystem in AI chips, software and developer adoption. The real issue is whether investors begin assigning serious value to alternative AI architectures.

Cerebras is not selling a normal chip story. It is selling scarcity, speed and a bet that AI workloads will become too large for one supplier to dominate forever.

That may be enough to get public-market attention. Whether it is enough to justify the price is the real test.