Key insights
- Data center construction delays due to power component shortages may curb AI infrastructure buildout. Dependence on Chinese electrical equipment and limited alternative suppliers from Canada, Mexico, and South Korea are bottlenecks. This could negatively impact demand for memory and GPUs in 2026, affecting semiconductor companies reliant on data center growth.
Companies like Amazon, Oracle, Meta, Google, and OpenAI have committed over 600B this year to building out datacenters. However, electrical power components like transformers, switchgears, and batteries, are not able to keep up with demand. Exports of these critical parts have increased coming out from China, but it isn't enoughto meet the unsatiable demand for AI in the US. US companies have also increased purchasing from companies from Canada, Mexico, and South Korea, but this is still not enough. Will this change the outlook on other supplier companies like memory and gpus for 2026?