Cerebras Systems is rapidly expanding the physical infrastructure supporting its AI inference business, with more than 600 megawatts of data center capacity now live or under contract for delivery by the end of 2027 and manufacturing capacity expected to increase more than 10x during 2026.
The AI infrastructure company has added new manufacturing lines at Flex, Sanmina and Rocket EMS to support the expansion. Cerebras also secured TSMC wafer supply that it says is sufficient to support its expected growth and is building a pipeline of potential data center opportunities measured in gigawatts.
Cerebras said its wafer-scale architecture provides supply-chain advantages as AI computing demand strains the availability of key semiconductor components. Its systems do not use HBM memory, CoWoS packaging or 3nm fabrication technology, which the company identified as currently supply constrained.
The infrastructure buildout follows a major expansion in financial capacity. Cerebras raised $6.4 billion in gross proceeds through its IPO and established an $850 million revolving credit facility intended to accelerate data center acquisitions. The company ended Q2 with $8.6 billion of cash, cash equivalents, restricted cash and short-term investments.
Remaining performance obligations reached $25.4 billion as of June 30, giving Cerebras substantial contracted business as it adds the capacity needed to fulfill future deployments. Management said it plans to more than triple revenue in 2027.
Cerebras’ cloud business is already scaling rapidly. Q2 GAAP cloud and other services revenue surged 281% year-over-year to $126 million, while core cloud and other services revenue increased 287% to $127.7 million. GAAP total revenue increased 74% to $180.1 million, and core revenue more than doubled to $209.9 million.
Core gross margin improved approximately 940 basis points to 41%, while core operating margin improved approximately 2,600 basis points to negative 16%. GAAP operating margin remained negative at 265%.
The company is also expanding its AI inference capabilities through major technology partnerships. Cerebras enabled OpenAI GPT-5.6 Sol at 750 tokens per second and developed a disaggregated inference solution with AMD that is expected to enter production in Q4 2026 and increase throughput by as much as 5x.
Cerebras expects to extend the same disaggregated architecture and potential 5x throughput benefit to Amazon Bedrock during the first quarter of 2027. New cloud capacity agreements were also signed with AI coding companies Cognition and Lovable, while customers using Cerebras inference for agentic workflows include Block, Figma, AlphaSense and GSK.
The company also highlighted a cybersecurity application with CrowdStrike, where Cerebras fast inference enables LLMs to operate inline across a large portion of enterprise traffic.
For Q3, Cerebras expects core revenue of approximately $214 million to $216 million and core gross margin of 38% to 40%. The company raised its full-year outlook to core revenue of $880 million to $890 million and a core gross margin of 41% to 43%.
KEY QUOTES:
“This was an outstanding quarter for Cerebras. Core revenue more than doubled to $210 million, and our cloud business nearly quadrupled year-over-year. Speed changes what AI can do. It makes AI more useful, more productive, and opens entirely new markets.”
“As a result, the demand for fast inference is enormous and Cerebras is scaling to meet it, securing more data center capacity, expanding manufacturing, and growing with customers and partners including OpenAI, AWS, AMD, and CrowdStrike.”
Andrew Feldman, Co-Founder And CEO Of Cerebras Systems
“Our quarterly results exceeded our guidance across all core business metrics. The market has responded strongly to the value of fast inference.”
“We have made rapid progress in key areas required to deliver exceptional growth against our remaining performance obligations of $25.4 billion, and plan to more than triple revenue in 2027.”
Bob Komin, Chief Financial Officer Of Cerebras Systems
