Why AI could create inflation before it lowers prices

The intuitive story is that AI makes work more efficient, reduces costs and lowers prices. But productivity gains usually appear only after companies reorganise processes, software and staffing. Before that comes an investment phase.

Data centres, chips, electricity, specialist workers and software can all be demanded at the same time. If supply cannot expand quickly enough, the investments that later improve efficiency may initially create price pressure.

That creates a difficult monetary-policy problem. A one-off investment surge is different from persistent inflation, but shortages can spread through wages, energy or semiconductors and affect many industries at once.

The SNB therefore has to measure not only whether AI raises productivity but when and through which channels. A long-run disinflationary effect can coexist with higher prices in the short run.

AI is no longer only a technology-sector story. If investment becomes large enough, it changes growth, capital demand, electricity use and potentially the transmission of monetary policy.

Tschudin’s point is mainly a warning against simple narratives. Technology can push prices down and up at different stages. For the SNB, measured inflation matters more than the promise of future efficiency.