BNP Paribas Equity Research estimates that annual AI hardware capital expenditure is approaching USD1 trillion, flowing across 30 segments of the data centre value chain. Chips and servers capture 50% of the total, power infrastructure 20%, networking equipment 15%, and cooling and facilities and construction 7.5% each. The report identifies critical bottlenecks including advanced logic foundry and packaging, DRAM and HBM, grid interconnection, lithography equipment, and indium phosphide optical transceivers. For a financial institution, this mapping provides a structured framework to evaluate which segments of the AI supply chain face the most severe physical and production constraints that could slow deployment timelines independent of demand.
300 other tracked developments touching the same claims.