Broadcom reported $16.7 billion in custom AI silicon revenue during its fiscal third quarter ended August 2, representing a 221% increase from the same period a year earlier and a 54% quarter-over-quarter gain.
Total company revenue reached $29.6 billion, up 86% year-over-year. The semiconductor solutions segment grew 127% to $20.8 billion, while infrastructure software climbed 29% to $8.75 billion.
CEO Hock Tan told investors the momentum is continuing, projecting AI semiconductor revenue will reach $21.7 billion in the fourth quarter—a 236% increase from the prior year period. Overall fourth-quarter revenue guidance stands at approximately $34.8 billion, up 93% year-over-year.
Broadcom's custom silicon strategy centers on application-specific integrated circuits (ASICs) designed for particular AI workloads. These chips are less flexible than general-purpose GPUs but offer cost advantages and performance tailored for inference tasks, which involve running trained models for end users.
Major customers building their own chips to reduce dependence on Nvidia GPUs include Google, which signed a long-term agreement with Broadcom through 2031 extending a roughly decade-long TPU partnership. OpenAI and Broadcom jointly unveiled an inference chip called Jalapeño in June, with both companies targeting 10 gigawatts of capacity eventually. Broadcom also committed to supplying Anthropic with 3.5 gigawatts of Google-powered computing capacity starting in 2027, potentially expanding to 5 gigawatts over time.
The company generated $13.7 billion in free cash flow during the quarter, representing approximately 46% of revenue, with a non-GAAP operating margin of 66%. The board declared a quarterly dividend of $0.65 per share.
