Industry Analysis
Cathie Wood’s exit from AMD and TSMC isn’t mere semiconductor skepticism—it’s a rejection of peaking mature-node economics and inflated geopolitical premiums. Technically, AI chips are racing below 3nm, yet TSMC’s 2nm GAA yields remain bottlenecked by EUV tooling, while AMD faces dual pressure from NVIDIA’s Grace Hopper and Intel’s Sierra Forest in data centers. Regulatory risks loom large: U.S. CHIPS Act subsidies tie production to domestic fabs, inflating TSMC Arizona’s costs by ~30%, while China’s 28nm self-sufficiency drive erodes TSMC’s mainland revenue. Competitively, NVIDIA will tighten its AI-stack dominance, and Samsung may undercut TSMC via HBM3E price wars. Over the next 12–24 months, capital will pivot from foundry capacity toward AI-native architectures, advanced packaging, and RISC-V ecosystems—Wood’s reallocation signals a bet on post-Moore valuation resets.
This page displays AI-generated summaries and metadata for research purposes. Original content belongs to the respective publishers.