Peter Schiff: The $1 Trillion AI CapEx Bubble Is Hiding a Massive Capital Misallocation

Peter Schiff argues on his podcast that hyperscalers’ roughly $1 trillion annual AI-related capex is a potential bubble and a massive capital misallocation, buying hardware that may become obsolete in 5–6 years. The article cites big capex figures: Microsoft $30.88B (Q1, +84% YoY), Alphabet $35.67B, Amazon $44.2B (quarter; annualized ~ $175B) and Meta raising 2026 capex guidance to $125–145B. Nvidia remains the primary beneficiary (Q1 FY27 revenue $81.62B, +85%), while Micron has surged on HBM demand. Market impact: stocks are being punished or rewarded based on whether capex converts to bookings — Microsoft and Meta lag YTD, while Google and Amazon are up. Schiff warns the spending could mask consumer weakness and crowd out other investments, while bulls point to accelerating AI revenues. The debate over depreciation timelines vs. revenue scaling will shape sector performance and investor positioning.