AI chipmaker Cerebras is gearing up for one of the biggest tech IPOs of the year, targeting a $3.5 billion raise that would value the company at $24.5 billion. That’s a notable bump from its $23 billion private valuation just three months ago in February, signaling strong investor appetite for alternatives to Nvidia’s dominance in AI infrastructure. The move comes as demand for specialized AI chips continues to surge across data centers worldwide.

Cerebras is making its big bet on public markets. The Sunnyvale-based AI chipmaker is targeting a $3.5 billion raise in what could become one of 2026’s most closely watched tech IPOs, according to reports from CNBC. The share sale would value the company at up to $24.5 billion, a 6.5% bump from its $23 billion valuation just three months earlier in February.

The timing is deliberate. AI infrastructure spending is hitting record levels as companies race to build out the computing power needed for large language models and generative AI applications. While Nvidia has captured most of the headlines and market share with its H100 and upcoming Blackwell GPUs, Cerebras has carved out a distinct position with its wafer-scale engine approach – essentially building chips the size of dinner plates that deliver massive parallel processing capabilities.

Cerebras’s IPO push comes at an interesting moment for the AI chip market. The company faces intense competition not just from Nvidia’s ecosystem but also from cloud giants like Amazon, Google, and Microsoft who are all developing custom AI silicon. Yet Cerebras has managed to attract notable customers including major pharmaceutical companies and research institutions that need extreme-scale AI training capabilities.

The valuation increase from $23 billion to $24.5 billion in just three months suggests strong investor confidence, but it also reflects the frothy valuations across AI infrastructure companies. The question for potential IPO investors will be whether Cerebras can demonstrate sustainable revenue growth and a clear path to profitability in a market where Nvidia controls an estimated 80-90% share of AI training chips.

What sets Cerebras apart is its technical architecture. The company’s CS-2 system uses a single wafer-scale chip containing 2.6 trillion transistors – compared to Nvidia’s H100 which has around 80 billion. This design allows for faster data movement and reduced bottlenecks during AI training, though it comes with manufacturing complexity and higher costs per unit.

The $3.5 billion raise would give Cerebras significant capital to scale production, expand its sales team, and invest in next-generation chip development. But the company will need to navigate supply chain challenges, convince more enterprise customers to bet on its platform, and prove it can coexist with or displace Nvidia installations in major data centers.

IPO market conditions have been choppy in 2026, with several tech companies delaying public debuts amid economic uncertainty. Cerebras’s decision to move forward suggests either strong demand from institutional investors or pressure from existing shareholders – including venture backers who’ve poured billions into the company since its 2016 founding – to create liquidity.

The AI chip sector has seen massive investment over the past two years, with startups raising billions to challenge Nvidia’s dominance. Most have struggled to gain meaningful market share. Cerebras will need to show not just impressive technology specs but actual customer adoption, recurring revenue, and unit economics that work at scale.

One key factor to watch in the S-1 filing will be customer concentration. If Cerebras derives a large percentage of revenue from just a handful of customers, that could raise red flags for public market investors. The company will also need to disclose its burn rate, gross margins, and how much of the IPO proceeds will fund operations versus providing exits for early investors.

The broader context is that AI infrastructure companies are getting premium valuations right now, but public market investors are increasingly sophisticated about separating hype from sustainable business models. Cerebras has genuine technology differentiation, but turning that into a profitable public company in a market dominated by Nvidia will be the ultimate test.

Cerebras’s $3.5 billion IPO at a $24.5 billion valuation will be a critical test for whether the public markets believe there’s room for credible alternatives to Nvidia in AI infrastructure. The 6.5% valuation bump since February shows momentum, but the real scrutiny begins when the S-1 drops and investors can see the actual revenue, margins, and customer base behind the impressive chip technology. For the broader AI ecosystem, a successful Cerebras IPO would validate that specialized architectures can compete with GPU incumbents and potentially open the door for other AI infrastructure startups eyeing public debuts. But if the offering stumbles, it could freeze the IPO market for AI hardware companies and reinforce Nvidia’s seemingly unassailable position.