Victoria Bazir
Inflation expectations are rising again while Treasury yields remain stuck near multi-year highs. That combination is starting to pressure one of the market’s most crowded trades: AI. Markets are no longer pricing a clean return to low-inflation conditions.
The U.S. 5-year breakeven inflation rate a market-based measure of expected inflation has climbed back toward 2.7% after falling below 2% in late 2024. The move is not dramatic yet, but it signals that investors are beginning to price inflation as persistent rather than temporary. Bond markets are reinforcing the same message.
The U.S. 10-year Treasury yield remains above 4.4%, far above the sub-2% environment that supported much of the previous growth-stock cycle. Elevated yields increase financing costs across the economy and reduce the attractiveness of long-duration assets whose valuations depend heavily on future earnings. That becomes especially relevant for AI.
The current AI expansion is not just a software story. It is an infrastructure cycle requiring enormous upfront capital: data centers, GPUs, semiconductor fabs, cooling systems, electricity contracts, and grid upgrades. As inflation expectations rise and yields stay elevated, the cost of building that infrastructure rises with them. This is where the market narrative begins to shift.
Investors treated AI as structurally insulated from macro pressure. But higher rates change how markets value future growth. Companies relying on aggressive expansion and long-term projections become more sensitive to capital costs and financing conditions.
The tension is becoming harder to ignore because the AI boom itself is increasing demand for power and industrial infrastructure. Utilities and grid operators are already preparing for a surge in electricity consumption tied to hyperscale AI deployment. The long-term AI thesis may still be deflationary. But the buildout phase looks inflationary, capital-intensive, and increasingly exposed to macro conditions. That does not mean the AI trade breaks overnight.
Victoria Bazir
