Federal Reserve Chair Kevin Warsh, in testimony to the Senate Banking Committee on 15 July, delivered a dual message: the surge in artificial-intelligence investment is pushing up measured prices, but he views this as a supply-side adjustment that need not become entrenched inflation. He also expressed dissatisfaction with the central bank’s current inflation metrics, calling them “imperfect” for capturing underlying price trends. The remarks, coming days after a benign US consumer inflation print, signal that the Fed is grappling with how to interpret the price effects of the AI boom and may be reluctant to tighten policy solely in response to tech-driven cost increases.

Warsh argued that AI represents a supply shock—unlike a geopolitical conflict that constrains supply—because it can elicit a supply response that tempers price rises. “I don’t see a one-off change in prices as necessarily inflationary,” he said, adding that it would be up to the Fed to judge. This stance aligns with the view from some market participants, such as Brazil’s Reach Capital, that the current capex boom in data centres and chips will pressure inflation before delivering productivity gains later this decade. However, the Bank for International Settlements has warned that the same investment surge, now exceeding $1.2 trillion since early 2025, could turn into a bust if returns disappoint, with cascading effects on financial conditions.

The AI spending spree has reshaped equity and debt markets. The so-called Magnificent Seven tech firms have lost $2 trillion in market value since October, even as their combined debt issuance topped $170 billion. Asian chipmakers—Taiwan’s TSMC, South Korea’s Samsung and SK Hynix—saw shares more than double in a half-year rally worth $1.8 trillion, but the Kospi index has since fallen 20% from its peak, with foreigners selling $100 billion in Korean equities. A survey by IDC across Asia-Pacific found that 37% of organisations are investing aggressively in AI with minimal assessment, yet only 40% say implementations have met or exceeded expectations. In Malaysia, 57.5% of firms invest based on AI’s potential without rigorous ROI analysis, and 80% of those whose projects underperformed reported higher-than-expected costs.

Warsh has given his newly formed policy review task forces six months to examine the Fed’s tools, including inflation measurement and the balance sheet. The next Federal Open Market Committee meeting is scheduled for 28–29 July, where officials will debate whether the current rate stance remains appropriate amid conflicting signals from AI-driven investment, a still-strong labour market, and uneven inflation data. Markets will watch for any shift in the Fed’s assessment of supply-side price dynamics and the risk that the AI investment cycle could amplify financial vulnerabilities.