Overview
- In late July federal officials and researchers sharpened their focus on big AI spending after Fed Chair Kevin Warsh told the Senate that the wave of investment represents real capital building productive capacity rather than a bubble.
- The Federal Reserve has formed internal task forces to study AI’s effects on jobs, productivity and monetary policy and Warsh said one‑time price jumps from GPUs and power should not be confused with persistent inflation.
- The Bank for International Settlements warned that the scale and timing of any productivity dividend are unclear and that heavy near‑term demand for data centers, chips and power creates a tradeoff for interest‑rate setters.
- Estimates from central bank analysts put combined hyperscaler capital commitments in the hundreds of billions to over $1 trillion across 2025–26, and those projects are already pushing up prices for GPUs, memory and electricity.
- The buildout raises financial‑stability questions because it draws on scarce labor and financing, big firms may be less sensitive to higher borrowing costs, and delayed productivity gains could leave lenders and local communities exposed to idle capacity or tighter grids.