AI Value Chain: What do we need to believe to believe the AI build is healthy?

This report focuses on the US Networking & Communications Equipment sector and the AI infrastructure value chain, examining the key assumptions one must believe for the AI build to be healthy. On the supply side, the report argues that AI infrastructure growth is well underpinned, with hyperscaler capex dollars continuing to expand; while there is little conviction in forecasting long-term AI end-user demand, the assumptions needed to underwrite continued investment (anchored on a roughly 15% ROI hurdle) appear plausible rather than aggressive. Technology leaders hold deep conviction in scaling compute, model capabilities continue to scale with compute and data, and AI adoption is outpacing past technology cycles. The report views model scaling and AI adoption as the key leading indicators, and both appear positive, leading it to net out positive on the AI infrastructure trade. Networking and optical companies are seen as structural winners as bandwidth increasingly becomes the bottlen