AI Infrastructure Debt Boom Raises New Risks for Institutional Investors
A record borrowing spree is fueling the rapid expansion of artificial‑intelligence (AI) data centers, according to a new analysis by former Treasury and Pentagon adviser Jim Rickards. In 2025, AI‑related companies raised at least $200 billion in debt, a figure that places the sector among the largest technology‑sector financing efforts in history.
The debt was issued through a mix of investment‑grade bonds, high‑yield debt and private‑credit arrangements. Many of these instruments are structured outside the traditional public markets that most investors monitor, creating a complex web of obligations that can be difficult to track.
Rickards’ presentation, now available for free, explains how the financing structures work and why the complexity itself may generate hidden risks. He points out that the sheer scale of the borrowing is unusual by historical standards and that the capital required to build the next generation of AI infrastructure could reach hundreds of billions of dollars over the next few years.
One key concern highlighted by analysts at JPMorgan is that as institutional investors increase their holdings of AI‑related debt, bond portfolios become more tied to the fortunes of technology companies than to traditional interest‑rate dynamics. The shift, Rickards says, carries risk beyond Silicon Valley and into pension funds, retirement accounts and diversified bond portfolios.
Oliver Wyman has also warned that lenders may discover they hold more exposure to data‑center and digital‑infrastructure risk than many internal models currently suggest. The firm’s observations reinforce the idea that the debt backing AI infrastructure is not a simple, isolated market.
Rickards stresses that he is not predicting a crisis. Instead, he urges investors to understand the financing mechanisms that support the AI buildout, especially because many ordinary investors may already be exposed to the debt through bond funds, retirement plans and institutional portfolios.
The upcoming earnings cycle on July 29, when several major AI‑linked companies are expected to report growth, spending and demand expectations, could serve as an important test of the assumptions underpinning the borrowing boom. If spending projections or demand forecasts soften, investors may begin to reassess the massive debt financing behind AI infrastructure.
The analysis is part of a broader conversation about how the AI revolution is reshaping capital markets. According to a recent report, AI infrastructure debt surged 112 % to $25 billion in 2025, driven by tech giants’ $75 billion in bonds for GPU and cloud projects. Other estimates suggest that at least $290 billion in debt financing has been secured for hyperscaler AI infrastructure initiatives across the industry.
While the data‑center build‑out is a critical component of AI development, the financing side raises questions about the long‑term sustainability of the sector. The risk is not limited to the companies that issue the debt; it extends to the investors who hold it.
Rickards’ free presentation is available through Paradigm Press, a well‑reviewed independent financial research publisher. The presentation aims to help viewers understand where the debt resides, how it is structured, and why it may matter to ordinary investors.
In summary, the AI infrastructure debt boom is a significant development that could have ripple effects across the broader financial system. Investors who hold AI‑related debt directly or indirectly should pay close attention to the July 29 earnings reports and to any changes in spending or demand forecasts that could influence the value of the debt.
The current situation remains that AI companies have raised at least $200 billion in debt in 2025, and that the debt is being financed through a mix of public and private instruments. The next steps for investors will be to monitor the July 29 earnings cycle and to reassess their exposure to AI infrastructure debt as new data becomes available.
