Broadcom Posts 143% AI Chip Revenue Jump; China Eyes Indium Export Curbs

Two signals from the AI chip supply chain: Broadcom's AI semiconductor business crossed $10B in a single quarter as China tightens scrutiny on indium, a material critical for next-generation data center optical interconnects.

📊 Broadcom's AI Chip Revenue Hits $10.8B, Up 143% Year Over Year

Decoded: Broadcom reported fiscal Q2 2026 revenue of $22.2 billion, up 48% year over year, with AI semiconductor revenue reaching a record $10.8 billion — a 143% increase from the same quarter a year prior. AI bookings exceeded $30 billion during the quarter, signaling sustained demand well into fiscal 2027. CEO Hock Tan said on the call that "the momentum continues," driven by hyperscaler orders for Broadcom's custom AI accelerators — XPUs — used in large-scale training and inference workloads. Broadcom's AI chip revenue now surpasses AMD's entire data center segment ($5.8B in Q1 2026) and exceeds Intel's Data Center and AI revenue by more than double. Broadcom also reported record operating profit and free cash flow for the quarter. (Broadcom Investor Relations, official Q2 FY2026 earnings release, June 2026)

Why it matters: Broadcom (AVGO) is consolidating its position as the primary custom AI silicon vendor for hyperscalers seeking alternatives to Nvidia's standard GPU model. Each of its three major XPU customers — Google, Meta, and a third undisclosed hyperscaler — represents a multi-billion dollar annual contract; $30B in quarterly bookings signals that the AI capex pipeline extends well beyond current fiscal year guidance. Unlike Nvidia's merchant GPU model, Broadcom's custom chip architecture gives hyperscalers differentiated performance per workload, a structural advantage as training and inference compute diverge. The 143% YoY growth rate also sets a credibility benchmark that AMD and Intel data center divisions will need to approach to remain competitive in 2027 hyperscaler procurement cycles.

🗄️ China Tightens Scrutiny of Indium Exports as AI Chip Demand Rises

Decoded: China has begun scrutinizing its exports of indium more closely, raising market concerns that formal export curbs could follow — a pattern consistent with Beijing's earlier moves on gallium and germanium. Indium phosphide, placed on China's export control list in February 2025, is a compound semiconductor material used in high-speed optical transceivers that connect GPU clusters inside AI data centers. Supply restrictions have become significant enough that the CEO of Coherent — an Nvidia-backed chipmaker manufacturing indium phosphide-based components — traveled to Beijing alongside a U.S. commercial delegation seeking supply clarity. China controls approximately 55-60% of global primary indium production, making it the dominant upstream supplier to Western chip manufacturers. (Reuters, June 19-20, 2026)

Why it matters: China's incremental tightening on indium follows the same escalation sequence used for gallium (restricted July 2023) and germanium (restricted August 2023), where heightened scrutiny preceded formal control orders by several months. Indium phosphide is not substitutable on short timelines — it enables the optical interconnects that determine GPU cluster bandwidth utilization, and sits upstream of every major AI chip vendor. Supply constraints here hit the AI data center build regardless of whether the end chips come from Nvidia, AMD, or Broadcom. Investors tracking AI infrastructure should monitor Coherent (COHR) and the pace of U.S. and Japanese efforts to develop ex-China indium refining capacity — a supply gap without a near-term industrial solution.

Stay decoded. See you tomorrow.

— The Get AI Decoded Team