Chip stocks race towards biggest gains since dotcom era on AI demand - Financial Times

Recent semiconductor stocks have surged to their largest gains since the dotcom era, driven primarily by the rapid advancement of artificial intelligence technology. The increased demand for high-perf... Read original →

The AI compute arms race is triggering a cascading effect across the semiconductor stack: NVIDIA’s GPU roadmap is forcing TSMC to accelerate 2nm and CoWoS advanced packaging capacity, while lithography, EDA, and HBM memory have become new bottlenecks. Geopolitical friction has translated into tangible compliance costs—U.S. export controls and the EU Chips Act compel supply chain reconfiguration. Foundries in Taiwan, China benefit from concentrated orders short-term but face long-term pressure to localize tech and diversify clients. In response to NVIDIA’s dominance in AI training chips, AMD, Intel, and Chinese players are countering with custom ASICs and chiplet-based asymmetric strategies. Over the next 18 months, capital will flood advanced manufacturing beyond mature nodes, risking structural overcapacity. The real tailwind lies not in chips alone, but in vertical integration from silicon to algorithms—reshaping the global semiconductor power structure.