Semiconductor Supply Chains

The global semiconductor supply chain that produces AI chips is remarkably concentrated and geographically constrained. The most advanced chips are designed primarily in the United States (NVIDIA, AMD, Apple) and fabricated almost exclusively by TSMC in Taiwan, using lithography machines made solely by ASML in the Netherlands. This concentration creates significant geopolitical risk. Tensions over Taiwan, export controls on advanced chips to China, and efforts by the US, EU, and other governments to build domestic semiconductor capacity have made chip supply chains a matter of national security. The US CHIPS Act and similar legislation in Europe and Japan are channelling billions into new fabrication plants, but building a cutting-edge fab takes years and costs upwards of $20 billion. For AI specifically, the supply chain bottleneck has been acute. During 2023 and 2024, demand for NVIDIA's top GPUs far exceeded supply, with wait times stretching to months. This imbalance has eased somewhat but continues to affect planning for large-scale AI projects. If you're making significant AI infrastructure investments, supply chain risk should factor into your decisions - including considering multi-vendor strategies and understanding the lead times involved in procuring hardware at scale.