The concerns flagged in the report cover a lot of ground. They include cyber threats, complex financing structures, and the possibility that an AI earnings downturn could drag down equity markets more broadly.
On-balance-sheet debt at the seven largest AI firms is projected to exceed $800 billion by the end of 2026.
Off-balance-sheet obligations tied to these firms exceed $3 trillion, including leases and guarantees.
Banks look fine, for now
Barr did offer some reassurance. Direct losses to banks from this AI-related financing appear manageable at present, he said, thanks to stable capital levels.
The surge in AI financing, both on and off balance sheets, raises concerns about higher market volatility and the potential for fire sales if conditions turn ugly.
Barr also pointed to hedge fund leverage as an amplifier. In his view, borrowed money at hedge funds could make a downturn in AI investment hit harder than it otherwise would.
The trading floor problem
Among the near-term risks he described were herding behavior in AI-driven trading and the possibility of market manipulation. If many firms lean on similar AI models, they may reach similar conclusions at the same moment.
Barr expressed cautious optimism about AI’s productivity benefits over the longer run.
A recurring theme for Barr
Barr has addressed both the risks and benefits of AI across multiple speeches throughout 2025 and 2026. He has previously raised similar points about elevated risk appetites and hedge fund leverage potentially worsening a sharp pullback in AI investment.
Equity valuations remain high, which adds to the financial stability concerns the Fed is tracking.
AI has also reshaped capital spending, particularly at major tech firms. That spending now sits at the intersection of inflation, monetary policy and market valuations.