Cerebras Systems experienced a sharp 14% drop in its stock price during extended trading following the release of its second quarterly earnings report since going public in May. Despite the market’s negative reaction, the artificial intelligence chipmaker raised its full-year core revenue guidance to between $880 million and $890 million, up from its previous forecast of $855 million to $865 million. The company’s leadership emphasized that demand for specialized AI hardware remains exceptionally strong.

For the second quarter, Cerebras reported core revenue of $180 million, falling short of the $194 million projected by analysts. Total revenue, which includes pass-through figures, reached $210 million. The company posted a net loss of $450.5 million for the quarter, a significant shift from the $309.5 million profit recorded in the same period last year. This loss was primarily driven by $386.6 million in stock-based compensation expenses. Looking ahead, Cerebras expects third-quarter core revenue to land between $214 million and $216 million, with gross margins expanding to the 38% to 40% range.

Positioned as a direct challenger to industry giant Nvidia, Cerebras specializes in low-latency “fast inference” chips designed to handle highly interactive AI workloads. CEO Andrew Feldman noted that these premium-priced chips are driving margin growth. The company also highlighted a massive backlog, reporting $25.4 billion in remaining performance obligations and projecting that its revenue will triple in the upcoming fiscal year.

Cerebras continues to expand its market footprint through strategic collaborations, including a partnership with Advanced Micro Devices (AMD) set to yield production-ready products later this year. Additionally, the company revealed that OpenAI is cleared to utilize its hardware to power the upcoming GPT 5.6 Sol model. Cerebras’s cloud division also showed strong momentum, generating $126 million in revenue during the quarter ending in June.

Key Takeaways

• Cerebras Systems raised its full-year core revenue guidance to $880M-$890M, but its stock fell 14% after missing Q2 revenue expectations.
• The chipmaker reported a net loss of $450.5 million, largely due to $386.6 million in stock-based compensation following its May IPO.
• Cerebras is targeting Nvidia's dominance in the AI space with high-margin 'fast inference' chips and has secured partnerships with AMD and OpenAI.

Editor’s Analysis & Impact

The post-earnings sell-off of Cerebras Systems highlights the intense scrutiny and high expectations currently facing the semiconductor sector. While a 14% drop reflects investor disappointment over a near-term revenue miss, the underlying fundamentals of Cerebras remain robust. The company’s focus on ‘fast inference’ addresses a critical bottleneck in AI deployment—latency—allowing it to command premium pricing and expand gross margins. Furthermore, securing partnerships with AMD and OpenAI validates Cerebras’s technology as a viable alternative to Nvidia’s dominant GPUs. With $25.4 billion in remaining performance obligations and plans to triple revenue next fiscal year, Cerebras is well-positioned for long-term scaling. However, the company must quickly transition from IPO-related stock compensation losses to sustained profitability to regain Wall Street’s confidence.

Frequently Asked Questions

Q: Why did Cerebras Systems' stock drop despite raising its full-year guidance?
A: The stock fell primarily because the company's Q2 core revenue of $180 million missed analyst expectations of $194 million, and it reported a substantial net loss due to stock-based compensation.

Q: What is 'fast inference' and why is it important for Cerebras?
A: Fast inference refers to the rapid processing of AI models to deliver low-latency, real-time responses. Cerebras specializes in these high-speed chips, which command premium prices and help boost the company's gross margins.

Q: Who are Cerebras's key partners in the AI industry?
A: Cerebras has partnered with Advanced Micro Devices (AMD) to integrate products later this year, and its chips are set to support OpenAI's upcoming GPT 5.6 Sol model.