Cerebras Goes Public: Why This AI Chip Company Could Challenge Nvidia

Celebras Systems, an AI chip startup that built one of the world's largest processors, just filed to go public. The company is positioning itself as a real challenger to Nvidia, which currently dominates the market for AI chips used in data centers.

What's the IPO About?

The Los Altos-based company submitted its registration paperwork to the SEC on April 18 and plans to list on Nasdaq under the ticker "CBRS." While the company hasn't revealed how much money it's raising or what it's worth, the timing is strategic—companies are spending heavily on AI infrastructure right now, and Cerebras wants to show investors it has a better way to do it than Nvidia.

A Different Approach to Building AI Chips

Cerebras' key innovation is its wafer-scale engine (WSE) architecture. Think of it this way: most AI chips are small and need to be connected together like puzzle pieces to handle complex tasks. Cerebras instead puts everything on one giant piece of silicon—an 8.5 by 8.5 inch wafer with 850,000 processing cores all connected internally.

This design has a major advantage: when processing cores need to talk to each other, they can do it instantly on the same chip instead of sending information across cables between separate chips. For training large language models (like ChatGPT), this faster communication can significantly speed up work and reduce energy waste.

How Well Is Cerebras Doing?

According to the filing, Cerebras made $78.4 million in 2025—a 230% increase from the year before. The company is selling its chips to pharmaceutical companies, government agencies, and cloud providers running large AI projects.

Note that these aren't mass-market sales. A single WSE system costs hundreds of thousands of dollars, so Cerebras targets customers with specific, demanding needs rather than trying to sell to everyone.

A Crowded But Growing Market

Cerebras isn't the only company trying to challenge Nvidia. Startups like Groq, SambaNova, and Graphcore are also building alternative AI chips. Meanwhile, tech giants Google, Amazon, and Microsoft are designing their own chips for their cloud services.

However, customers are increasingly interested in alternatives because Nvidia chips have been hard to get, expensive, and not always the best fit for every type of AI work. This creates opportunities for companies like Cerebras that excel at specific tasks—even if they can't beat Nvidia at everything.

The Manufacturing Challenge

Cerebras makes its chips through Taiwan Semiconductor Manufacturing Company using cutting-edge manufacturing techniques. Building a chip the size of an entire wafer is technically difficult—managing defects, delivering power, and keeping it cool all require specialized solutions.

This complexity is both good and bad. It's good because competitors can't easily copy the design, but it's bad because it makes it harder to scale up production quickly compared to traditional chip designs.

What Investors Need to Watch

The real question for public investors is whether Cerebras' technical advantages will hold up as AI becomes mainstream. Right now, the company benefits from GPU shortages and companies' willingness to try new approaches. But can it compete when supply improves and AI infrastructure becomes routine?

Cerebras will need to expand beyond its current customers while proving that its chips are better than Nvidia's for reasons that will matter long-term, not just because GPUs are hard to find right now.

What This Means for the AI Chip Industry

If Cerebras succeeds as a public company, it could inspire other startups with innovative chip designs to pursue IPOs and access capital. It would signal that public investors are willing to bet on alternatives to Nvidia, not just incremental improvements to GPU technology.

For the broader AI industry, Cerebras' success or failure will help answer an important question: Does innovation in AI acceleration require multiple competing approaches, or will Nvidia's existing advantages prove too difficult to overcome? The technical merits of Cerebras' approach are clear for certain tasks, but translating that into long-term business success is a different challenge entirely.