Nine tech giants carry $3 trillion in off-books AI commitments

A Wall Street Journal analysis of securities filings puts roughly $3 trillion in AI-related obligations on the books of just nine major technology companies, and almost none of it sits on their balance sheets. The figure covers financing structures that stretch beyond conventional corporate debt, raising questions about who actually bears the risk if AI spending slows.

A large slice of that total, between $904 billion and $1.2 trillion, consists of leases for data-center capacity that has been signed but not yet built. That gap between committed and constructed capacity is itself a source of uncertainty for anyone trying to size the AI infrastructure buildout.

Wharton economist João Gomes has flagged the financing mechanics behind these commitments as poorly understood, even by regulators. He points to a parallel funding system built on leverage and structures that sit outside traditional banking oversight.

Gomes warned that this fast-moving financing system, reliant on high leverage and structural weaknesses, has not been fully grasped by the Federal Reserve.

The Journal's reporting adds that private capital flows now financing AI infrastructure are increasingly bypassing the banking system altogether. That shift makes it harder for regulators to trace exposure across lenders, landlords, and equipment vendors tied to AI data centers.

Inside the Federal Reserve, Chair Kevin Warsh is reportedly weighing two competing narratives: whether AI investment is driving genuine productivity gains, or whether it is generating inflationary and financial pressures that could destabilize markets before those productivity gains materialize. The two outcomes call for very different policy responses, and the Fed's read on which is closer to reality will shape how aggressively it watches this off-balance-sheet buildup.

Morgan Stanley's own projections give a sense of scale for what's coming. The bank estimates nearly $3 trillion in global AI infrastructure spending through 2028, with roughly $1.5 trillion of that needing to come from external financing rather than corporate cash flow.

That financing gap is the crux of the concern. If AI revenue growth fails to keep pace with the capital being deployed, the debt and lease structures underpinning today's data-center expansion could become a stress point for lenders and investors who currently have limited visibility into how interconnected these obligations really are.