Broadcom Earnings: Record AI Revenue Hits $10.8B as Stock Slips on Software Miss

Broadcom Inc. delivered a record second-quarter performance with AI chip revenue hitting $10.8 billion, a 143% increase from the same period last year and exceeding its own forecast of $10.7 billion. The company reported total revenue of $22.19 billion, up 48% year-over-year, with adjusted earnings per share at $2.44, surpassing Wall Street expectations of $2.40. Free cash flow reached $10.26 billion, representing 46% of revenue. Despite the AI revenue beat, shares fell roughly 3% in after-hours trading due to a narrow miss on total revenue ($80 million short of analyst projections) and a more significant shortfall in infrastructure software revenue, which came in at $7.18 billion versus expectations of $7.32 billion. CEO Hock Tan attributed the AI revenue growth to accelerating demand for custom AI accelerators and AI networking solutions. For Q3, Broadcom guided AI semiconductor revenue to approximately $16 billion, a more than 200% year-over-year increase, and total revenue to around $29.4 billion, an 84% rise from the prior year. Analysts had projected total revenue of roughly $28.53 billion. The company’s AI revenue includes about 40% from networking silicon, distinguishing its strategy from competitors like Nvidia, which focuses on general-purpose GPUs. Broadcom’s AI business relies on co-designing application-specific integrated circuits (ASICs), called XPUs, tailored to individual hyperscaler customers like Google, Meta, and OpenAI. This process involves embedding Broadcom engineers within client teams for 18 to 24 months to optimize chips for specific workloads, such as Google’s Gemini inference stack or Meta’s recommendation systems. Fabricated on TSMC’s 3-nanometer process, these XPUs deliver superior performance-per-watt ratios for targeted tasks, with cost-of-ownership advantages estimated at 30-50% for production inference at hyperscaler scale. The infrastructure software miss, particularly in VMware, contributed to investor caution despite the AI revenue surge. Bernstein analyst Stacy Rasgon noted that forward AI guidance, rather than the headline beat, was driving the stock’s decline. Broadcom’s strategy contrasts sharply with Nvidia’s general-purpose approach, emphasizing customization and efficiency for hyperscale clients. The company’s Q2 results highlight the growing demand for specialized AI hardware amid rapid industry expansion.