Cerebras Systems just hit pause on its IPO dreams, but CEO Andrew Feldman insists it's not giving up on going public. The AI chipmaker withdrew its registration Friday after raising $1.1 billion at an $8.1 billion valuation, saying it needs to update investors on how much the business has changed since filing a year ago. It's an unusual move that highlights how fast the AI chip landscape is shifting.
The AI chip wars just got messier. Cerebras Systems, the startup billing itself as Nvidia's biggest threat in AI computing, pulled a surprise move Friday by withdrawing its IPO registration - then spent the weekend explaining why.
CEO Andrew Feldman broke his silence Sunday night with a LinkedIn post admitting the company "made a mistake" by not immediately explaining the withdrawal. His explanation reveals just how dramatically Cerebras has evolved since first filing to go public a year ago.
"Given that the business has improved in meaningful ways we decided to withdraw so that we can re-file with updated financials, strategy information including our approach to this rapidly changing AI landscape," Feldman wrote. The timing couldn't be more telling - days before pulling the IPO, Cerebras announced a massive $1.1 billion funding round at an $8.1 billion valuation.
That funding round brought in heavyweight investors including Tiger Global and 1789 Capital, where Donald Trump Jr. serves as a partner. Neither firm was mentioned in Cerebras' original 2024 filing, highlighting how much has changed in the company's investor base and strategic positioning.
The withdrawal comes as Cerebras has fundamentally shifted its business model. Originally positioned as a chip manufacturer competing with Nvidia's graphics processing units, the company has pivoted hard into cloud services. It now operates data centers that can field AI model requests directly, essentially becoming both the hardware maker and the service provider.
This evolution puts Cerebras in direct competition with cloud giants while still battling Nvidia for chip supremacy. The company continues to claim its wafer-scale processors outperform traditional GPUs for AI training and inference workloads. But the market dynamics have shifted dramatically since its initial filing.
Just Monday, AMD announced that OpenAI committed to potentially 6 gigawatts worth of its AI processors, with the possibility of owning up to 10% of the chipmaker. That deal underscores how aggressively companies are seeking alternatives to Nvidia's dominant position in AI hardware.
