Cerebras Systems posted its first quarterly results as a public company on Tuesday, and the top-line numbers are genuinely strong: core revenue of $191.3 million in Q1 2026, up 92% year-on-year from $99.5 million. GAAP revenue came in at $193.4 million, up 94%. Hardware revenue grew 60% to $111.6 million. Cloud and services revenue grew 167% to $79.8 million. Net loss narrowed from $23.9 million to $14 million. The company raised its full-year guidance, projecting core revenue of $855 to $865 million, representing 69% growth at the midpoint. And the stock fell 10% in extended trading. The explanation for that gap between strong revenue and a declining stock price is the gross margin guidance for Q2, which is what Cerebras itself announced alongside the beat and the one figure that NewsTrackerToday anchors as the number that explains the market’s reaction.
Core gross margin came in at 47% for Q1. For Q2, Cerebras guided to 36% to 38%. That is a compression of nine to eleven percentage points in a single quarter. The company’s explanation is straightforward: it is shifting revenue mix toward cloud deployments to fulfill large contract commitments, including the $20 billion multi-year agreement with OpenAI for 750 megawatts of Cerebras compute. Hardware sales, which carry higher gross margins in the current period, will decline for the next several quarters as that production transitions into the Cerebras cloud. Cloud contracts generate recurring revenue but at lower near-term gross margin than direct chip sales. The wafer-scale chip architecture that differentiates Cerebras from GPU-based alternatives – the WSE-3 chip is the size of an entire silicon wafer rather than a small die – is what drives the speed advantage the company cites against Nvidia and Broadcom in AI inference workloads.
Sophie Leclerc, who covers the technology sector, reads the architecture and the market position carefully: “Cerebras’ wafer-scale approach genuinely delivers faster inference for specific workloads – the throughput numbers on large language model inference are real and documented. The question I keep returning to is whether ‘faster inference’ is the buying criterion that determines where enterprise AI infrastructure capital flows, or whether ‘compatible with my existing stack’ and ‘backed by Nvidia’s ecosystem’ are more powerful purchasing motivations for most buyers. The OpenAI deal and the AWS partnership are the two data points that suggest Cerebras has cleared the credibility threshold. What margin compression says is that clearing that threshold comes at a cost.” The trajectory of cloud gross margins as scale increases is what NewsTrackerToday tracks: Cerebras has guided to 36-38% in Q2, which the company frames as transitional.
Liam Anderson reads the IPO math: “IPO at $185. Open at $350. Current price around $226 after the 10% post-earnings drop. The market loved the chip architecture story at $350. The market is repricing toward what the actual financials say at $226. A 47% gross margin in Q1 compressing to 37% in Q2 is a real number, not a noise event. The company’s argument that margins recover as cloud revenue scales needs to show up in Q3 data for the current valuation to hold.” Cerebras also disclosed a $24.6 billion remaining performance obligation backlog, meaning contracted future revenue that has not yet been recognized. The OpenAI deal’s 750 megawatts of compute commitment and the AWS partnership to deploy Cerebras systems in AWS data centers are the two structural anchors of that backlog. Neither is speculative in the way that pipeline revenue typically is.
The concentration risk in the Cerebras customer base is the fact that the company’s investors are watching most carefully alongside the gross margin story. In 2025, approximately 86% of revenue came from UAE-affiliated customers. Q1 2026 showed geographic diversification through the OpenAI and AWS relationships, but the shift from a single concentrated customer base to a diversified one is what NewsTrackerToday stays with as the underlying business quality question: a company growing at 92% with $24.6 billion in backlog but historically concentrated in one geography and now transitioning to a different revenue mix is a compelling investment thesis with a specific execution risk profile.
So does the gross margin compression represent a temporary transition to a more durable cloud revenue model, or does it signal a structural challenge in Cerebras’ ability to maintain pricing power against Nvidia’s ecosystem as hyperscalers deploy more of their own custom ASICs? The Q3 gross margin print, which will be the first quarter with the full effect of the cloud revenue ramp in place, is the data point that resolves that question in one direction or the other. That is the specific number the company has effectively put on trial, and it is what News Tracker Today closes on as the test the next earnings cycle will run against the infrastructure-play valuation Cerebras commanded at its $350 IPO debut.
