Broadcom (AVGO) said it faces supply chain constraints including capacity limits at manufacturing partner TSMC, highlighting ripple effects of soaring AI chip demand across the broader tech industry.
The capacity warnings signal that even well-positioned chipmakers are struggling to secure sufficient manufacturing slots, potentially impacting revenue growth and customer delivery timelines.
Key Takeaways
- TSMC tells Nvidia and Broadcom it cannot meet full demand
- AI chip capacity constraints emerge as industry-wide bottleneck
- Intel positioned as alternative supplier amid supply crunch
Market Context and Supply Chain Pressure
Taiwan Semiconductor Manufacturing Company (TSMC), the world’s largest contract chipmaker, has informed key customers including Nvidia (NVDA) and Broadcom that it cannot meet all of their growing demand for advanced AI processors 1. The development underscores mounting pressure on the global semiconductor supply chain as AI-driven demand accelerates beyond manufacturing capacity.
TSMC’s advanced process nodes, particularly 3nm and below, are running near full utilization with limited flexibility even for top-tier clients. The capacity constraints affect the broader AI infrastructure buildout, where cutting-edge manufacturing capacity remains severely limited despite surging demand from hyperscale cloud providers and chip designers.
Broadcom’s Strategic Response
Despite supply chain challenges, Broadcom has positioned itself strategically within the AI chip ecosystem through application-specific integrated circuit (ASIC) development. Major technology companies including Google, Meta, Microsoft, and Amazon are working with Broadcom to develop custom AI chips tailored to their specific AI models, reducing dependence on standard GPU offerings 2.
The company reported first-quarter fiscal 2026 revenue of $19.31 billion, representing 29% year-over-year growth, with AI revenue reaching $8.4 billion, up 106% from the previous year. CEO Hock Tan projected the company could achieve more than $100 billion in AI revenue next year in the semiconductor sector alone 3.
Industry Implications and Alternative Suppliers
The capacity constraints at TSMC are creating opportunities for alternative foundry providers, particularly Intel’s foundry services. Intel is positioning itself as a key player in alleviating supply chain pressures through available capacity and geographic diversification, which becomes crucial for customers facing multi-quarter delays 4.
According to industry analysis, the situation reflects a broader shift where reliability and availability may outweigh the need for best-in-class performance. Kim Se-hwan, an analyst at KB Securities, said the results suggest “the AI infrastructure market is expanding from a GPU-dominated structure to ASIC-based accelerators” as customers seek alternatives to address supply bottlenecks and high total cost of ownership 5.
Long-term Market Dynamics
The supply constraints highlight a fundamental shift in the AI chip market structure. While demand validation is no longer in question, the focus has moved to capacity allocation and supply chain security.
TSMC’s capacity limits do not diminish the AI boom but rather affirm its sustained growth potential. The situation may accelerate diversification strategies among major technology companies and provide renewed relevance for Intel’s foundry expansion efforts as the industry seeks to reduce dependence on a single manufacturing hub.
Not investment advice. For informational purposes only.
References
1Dev Kundaliya (January 19, 2026). “TSMC warns Nvidia and Broadcom of capacity squeeze as AI chip demand surges”. Computing. Retrieved March 24, 2026.
2AInvest (March 6, 2026). “Broadcom’s $100B AI Chip Play Hinges on TSMC’s 3nm Execution Window”. AInvest. Retrieved March 24, 2026.
3Yiannis Zourmpanos (March 15, 2026). “TSMC and the AI Bottleneck: Why the Thesis Is Maturing, Not Breaking”. Yahoo Finance. Retrieved March 24, 2026.
4Emily J. Thompson (January 14, 2026). “TSMC’s Capacity Constraints Impact AI Chip Supply”. Intellectia.AI. Retrieved March 24, 2026.
5Chun Byung-soo (March 6, 2026). “Broadcom locks HBM and TSMC capacity, challenges Nvidia and AMD in AI”. CHOSUNBIZ. Retrieved March 24, 2026.