The $5 Trillion Question | Kai Wu on the Risks of the Mag Seven's Big AI CapEx Bet

Kai Wu of Sparkline Capital joins Excess Returns to discuss his paper Surviving the AI CapEx Boom. In this episode, Kai breaks down the unprecedented level of investment in AI infrastructure, why today’s AI buildout mirrors past technology booms, and what it all means for investors. He explores the parallels between AI and historic bubbles, the implications of massive corporate CapEx spending, and where value might ultimately be captured as the cycle plays out.Topics covered:Why big tech’s CapEx spending has exploded and how much they’re investingThe trillions in revenue needed to justify AI infrastructure spendingHistorical parallels with the railroad and dot-com buildoutsWhy companies that invest heavily often underperformHow the Mag 7 are shifting from asset-light to asset-heavy businessesThe risks of “circular deals” and financial entanglement in AIWhy the AI race resembles a prisoner’s dilemmaWhich layers of the AI stack may capture long-term valueHow early adopters and infrastructure players differ in capital intensity and returnsWhere investors might find opportunity beyond the obvious AI namesTimestamps:00:00 Introduction and overview of AI CapEx boom03:00 Why Kai researched AI investment cycles05:00 Scale of big tech’s CapEx spending07:00 Revenue needed to justify AI infrastructure08:30 Market concentration and valuation risks11:30 Historical parallels: railroads, internet, and AI14:30 The capital cycle and overinvestment dynamics17:30 “This time is different?” and lessons from bubbles18:00 Factor investing and high-asset-growth underperformance21:00 Sector and firm-level CapEx trends22:30 Winner-take-all dynamics and competitive pressure26:00 How the Mag 7’s business model is changing30:00 Comparing tech CapEx to utilities34:00 The circular deal problem and financial risk37:30 The AI arms race as a prisoner’s dilemma40:30 Will AI be winner-take-all?43:30 Lessons from the railroad and dot-com eras47:00 Where the value is captured in infrastructure vs adoption48:00 Identifying early AI adopters and hidden beneficiaries50:30 Sector and geographic AI exposure54:00 Capital intensity and valuation differences between infrastructure and adopters