Cerebras Systems, the AI chip company that pulled off the largest US tech IPO of 2026, just filed its first-ever quarterly earnings report. Revenue nearly doubled. The net loss shrank. And the stock dropped anyway.

The company reported Q1 2026 revenue of $193.4 million on June 23, representing 92% year-over-year growth from $99.5 million in the same period last year. That number actually beat analyst expectations, which had clustered around $180 to $183 million. Shares fell roughly 8% in after-hours trading, with investors fixating instead on what the company signaled about its margins going forward.

The numbers look good on paper

Cerebras narrowed its net loss to $14 million, or $0.22 per share, down from a $23.9 million loss a year earlier. Analysts had expected a loss of $0.14 to $0.16 per share, so the company missed on the bottom line even as it beat on the top line.

The revenue trajectory is genuinely striking. The company pulled in just $24.6 million in 2022. By 2025, that figure had ballooned to $510 million, a 76% year-over-year jump. For full-year 2025, Cerebras actually swung to a GAAP net income of $237.8 million from a $481.6 million loss in 2024.

Cerebras went public on May 14, 2026, pricing shares at $185 and raising approximately $5.5 billion in the process. On its first day of trading, shares rocketed as high as $350, an 89% premium, before settling back to close at $311, still up 68% from the IPO price.

This earnings report landed just five weeks after the company started trading publicly.

Why investors hit the sell button

The sell-off centers on margins. Cerebras’ forward guidance on gross margins gave investors pause. This dynamic is especially pronounced for AI hardware companies. Each Cerebras wafer-scale engine is roughly 56 times larger than a typical GPU chip. That manufacturing approach unlocks performance advantages for certain AI workloads, but it also creates a cost structure that is difficult to optimize.

Nvidia’s data center GPU business operates with gross margins that have at times exceeded 70%. Any AI hardware competitor needs to demonstrate a credible path toward similar profitability.

What this means for investors watching AI infrastructure

Cerebras occupies an interesting position in the AI hardware landscape. It’s not competing head-to-head with Nvidia for general-purpose GPU dominance. Instead, it’s carving out a niche in purpose-built AI training and inference hardware, targeting workloads where its wafer-scale architecture offers a genuine technical edge.

Cerebras has historically derived significant revenue from a small number of large customers. The broader takeaway for the AI sector is that revenue growth alone is no longer enough to sustain premium valuations. Cerebras’ first earnings report — strong on growth, shaky on margins — is a case study in that transition.