Nvidia Rival Cerebras (CBRS.US) Hits Record High Q3 Revenue, Cloud Business Income Surges 265% Year-on-Year

According to Zhitong Finance APP, artificial intelligence chip manufacturer Cerebras Systems (CBRS.US) released its financial report for the third quarter of 2026 on Thursday. Driven by strong demand for high-speed AI inference, the company achieved a record quarterly revenue and raised its full-year performance guidance. Meanwhile, the company's Q4 revenue outlook far exceeded Wall Street expectations, indicating further breakthroughs in the Nvidia (NVDA.US)-dominated AI chip market. Following the earnings release, Cerebras’ share price rose more than 15% in after-hours trading. As of Thursday’s close, the stock had fallen 47% year-to-date.

The financial report shows that Cerebras’ third quarter revenue calculated under US GAAP reached $210.8 million, an increase of 55% year-on-year, setting a new record high. On a non-GAAP basis, the company's core revenue for the third quarter reached $230.9 million, a 70% increase year-on-year, also reaching a record high.

In terms of profitability, the company’s GAAP net loss for Q3 was $161.2 million, significantly higher than $22.2 million in the same period last year; the diluted loss per share was $0.67, compared to $0.41 a year earlier.

However, on a non-GAAP basis, the company achieved core net profit of $29.82 million in Q3, compared to a net loss of $10.86 million in the same period last year, turning profitable on a year-on-year basis.

The company’s GAAP gross margin for Q3 was 20%, down from 40.6% in the same period last year but improved from the previous quarter. Core gross margin was 41%, higher than the 40.7% recorded in the same period last year. Q3 GAAP operating loss was $224.1 million versus $26.47 million the previous year; core operating loss was $33.11 million, compared to $15.12 million last year. Core operating margin was negative 14.3%.

Notably, stock-based compensation and customer warrant amortization were important reasons for the significant difference between GAAP and core results. The company’s Q3 stock-based compensation reached $155.1 million and customer warrant amortization was $36.42 million.

Q4 Revenue Guidance Far Exceeds Expectations, Revenue Expected to Triple by 2027

Looking ahead to Q4, Cerebras expects non-GAAP core revenue of $310 million to $330 million, with a midpoint of $320 million, approximately 18% higher than the analyst consensus of $271.2 million compiled by the media.

Based on the midpoint guidance, the company’s Q4 core revenue is expected to grow by approximately 39% from Q3. Cerebras also projects that Q4 core gross margin will be between 47% and 49%, up from 41% in Q3; core operating margin will range from negative 1% to positive 1%, indicating the core business could approach break-even in Q4.

For the full year 2026, Cerebras raised its outlook, forecasting core revenue of $940 million to $960 million, core gross margin of 42% to 44%, and core operating margin of negative 10% to negative 8%.

Cerebras co-founder and CEO Andrew Feldman stated the company is providing more channels for customers to use its technology through direct hardware system sales, its own cloud platform, and third-party cloud service providers, expecting 2027 revenue to reach triple that of 2026.

CFO Bob Komin noted the company has exceeded its own guidance for three consecutive quarters, and that all core financial indicators improved sequentially in Q3. Cerebras also won important customers including major cloud service providers, emerging AI cloud computing companies, and large enterprises, with potential opportunities in high-speed AI inference expanding to more than 10 times the previous scale during the quarter.

Cloud Business Revenue Surges 265% YOY, Becomes Main Growth Engine

Cerebras currently has two main business lines: selling computing systems equipped with self-developed AI chips, and providing cloud computing services based on its chip technology.

By business segment, the company’s Q3 GAAP cloud and other services revenue grew 265% year-on-year to $141.9 million, a record high, accounting for about 67% of total revenue. On a non-GAAP basis, core cloud and other services revenue reached $144.9 million, up 272% year-on-year. This segment’s core gross margin jumped from 16.5% last year to 46.9%.

By contrast, the hardware business came under some pressure. Q3 GAAP hardware revenue was $68.88 million, down about 29% from $96.79 million last year; core hardware revenue was $85.96 million, down about 11% year-on-year. Hardware GAAP gross margin fell from 50.4% last year to 11%, while core gross margin declined from 50.5% to 31%. The rapid expansion of the cloud business indicates that Cerebras is shifting from primarily relying on AI hardware system sales to a dual model of hardware sales and parallel cloud computing services.

Secures Orders from Two Major Cloud Providers, Remaining Performance Obligations Reach $26.1 Billion

In customer expansion, Cerebras disclosed that it achieved business cooperation with two new hyperscale cloud service providers in Q3. Among these, Oracle (ORCL.US) is integrating Cerebras’ high-speed AI inference technology into its cloud infrastructure platform (OCI), with services expected to launch in 2027, providing enterprise-grade inference capability to AI labs, enterprises, and developers.

Additionally, Cerebras announced a major cloud computing partnership with one of the world’s top three cloud providers, with further details expected to be disclosed next year.

The company is also expanding its technology reach by collaborating with emerging AI cloud service providers, reducing dependence on any single sales channel. As of the end of September, Cerebras’ total remaining performance obligations reached $26.1 billion, reflecting the future revenue from already-signed contracts.

In order to meet the growing demand for AI computing power, Cerebras has secured 700 MW of data center capacity, which is expected to be delivered by the end of 2027. The company has also signed additional data center capacity agreements in Norway and Finland to expand its European presence.

On the manufacturing side, Cerebras is collaborating with Flex and Sanmina to expand production capacity, expecting manufacturing capacity to exceed ten times the beginning-of-2026 level by year-end and planning further plant expansion in 2027.

Accelerating Challenge to Nvidia, AI Chip Competition Enters New Stage

Cerebras, which completed its IPO in May this year, is seeking to expand its share of the Nvidia-dominated AI chip market and benefit from the global data center infrastructure investment boom through its unique wafer-scale chip design.

Unlike traditional GPU designs, Cerebras uses wafer-scale processor architecture, integrating a massive number of compute units on a single wafer to deliver high-performance computing for AI models.

In August, Cerebras launched the next-generation AI computing system CS-4. The company claims this system can achieve up to 30x speed improvement in inference and will soon enter full production.

Meanwhile, the company also showcased its plan to release the next-generation computing system CS-5 in early summer 2027, aiming for up to 10,000 AI tokens generated per user per second. The system’s new generation wafer-scale processor WSE-5 has completed preliminary debugging and is currently undergoing system testing.

Cerebras has also introduced a heterogeneous disaggregated inference solution for production environments, optimizing compute resource allocation across different inference stages to improve power efficiency in data centers and reduce costs for AI users.

Cerebras’ strong Q4 outlook sends a positive signal for its strategy to promote new chip architectures and challenge Nvidia’s market leadership. At the same time, Nvidia is facing uncertainty due to executive changes. On Thursday, Nvidia announced the dismissal of Jensen Huang from the CEO role, a sudden event that has also drawn more attention to the competitive landscape of the AI chip industry.