定制芯片，两大赢家

Core Insights - Broadcom and TSMC are major beneficiaries of the custom AI chip trend, alongside Nvidia's continued dominance in the AI infrastructure market [1][2] - Broadcom reported Q1 FY2026 revenue of $18.02 billion, a 28% year-over-year increase, with AI semiconductor revenue soaring by 74% [1] - TSMC's Q4 FY2025 revenue reached $33.73 billion, a 20.5% year-over-year increase, with 77% of wafer revenue coming from advanced processes of 7nm and below [1] Group 1 - Broadcom's AI semiconductor revenue is expected to double to $8.2 billion in Q1 FY2026, driven by custom AI accelerators and Ethernet AI switches [1] - Broadcom's profit growth reached 188%, with a stable gross margin of around 68%, showcasing strong pricing power in proprietary chip architecture for large data centers [1] - TSMC's 3nm process accounted for 28% of wafer revenue, while the 5nm process represented 35%, indicating strong demand for cutting-edge AI chip manufacturing [1] Group 2 - The trend of custom chips signifies a strategic shift, with companies like Alphabet, Amazon, and Meta Platforms designing specialized accelerators rather than relying solely on Nvidia's GPUs [2] - TSMC expects Q1 FY2026 revenue to be between $34.6 billion and $35.8 billion, while Broadcom anticipates $19.1 billion [2] - The growth of custom chips will not replace Nvidia's dominance but will create new revenue streams benefiting the entire semiconductor ecosystem [2]