TSMC Risk

This case study examines TSMC's role as a bottleneck in the AI industry boom and the strategic implications of supply chain dependencies. The key challenge is that while Big Tech companies (Amazon, Microsoft, Google, Meta) are experiencing massive demand for AI infrastructure that exceeds their supply capacity, the constraint isn't power or data centers—it's semiconductor production capacity at TSMC.

The core issue stems from TSMC's conservative investment approach following the AI revolution. While ChatGPT launched in November 2022 and triggered massive CapEx increases among hyperscalers, TSMC's capital expenditures remained flat and even declined in 2023-2024. This created a significant supply-demand imbalance, as semiconductor fabs take 2-3 years to build. TSMC finally increased CapEx by 37% to $41 billion in the most recent year, with plans for $52-56 billion this year, but these investments won't meaningfully impact supply until 2028-2029.

The outcome is quantifiable revenue loss across Big Tech, as companies explicitly stated they couldn't monetize existing demand due to chip shortages. TSMC CEO C.C. Wei acknowledged that "silicon from TSMC is a bottleneck" according to customers.

**Key PM Takeaway**: In platform-dependent ecosystems, the most conservative player often determines the growth ceiling for the entire industry. Product managers must identify critical supply chain bottlenecks early and work to diversify dependencies or influence capacity planning of key suppliers before demand spikes occur.