The AI Infrastructure Build-Out Faces a $1 Trillion Funding Gap

The sums required to build out artificial-intelligence infrastructure are so vast that the traditional bond market may not be able to finance them in full. According to a Friday note from Torsten Slok, chief economist at Apollo Global Management, that gap could run to roughly $1 trillion — and it will likely have to be filled through private credit, debt provided not by banks but by specialized funds.

AI-related borrowing now accounts for more than 40% of new long-term, investment-grade corporate debt — that is, debt carrying the highest, most reliable credit ratings, Slok writes. Through 2030, the investment-grade market can absorb only a portion of the required amount — less than $1 trillion — because of "concentration and rating constraints": investors may already have committed heavily to the companies leading the AI build-out.

That gap, Slok says, can be filled by private lenders and other forms of financing backed by infrastructure, equipment, and individual AI projects. Unlike traditional unsecured corporate bonds, such private deals offer lenders more protection by tying the debt to specific assets or contractual guarantees.

The figures make the scale plain: among the "Magnificent Seven" — the seven largest technology companies leading the US stock market — Amazon has the highest capital-expenditure forecast at $220 billion, followed by Alphabet ($205 billion) and Microsoft ($175 billion). Together, Amazon, Alphabet, Microsoft, and Meta plan to spend $738 billion in the current fiscal year.

And companies keep expanding their plans: Nvidia and OpenAI are reportedly discussing a data center near Columbus, Ohio, whose cost could top $500 billion — including chip purchases worth as much as $350 billion. If it goes ahead, the project could become the largest of its kind, pointing to even greater demand for AI-infrastructure financing.