Not just the hottest stock market concept, AI has already become a major story in the U.S. bond market

The AI boom is spreading from the stock market to the bond market, with the debt scale of AI-related companies surpassing that of traditional banking, becoming the largest sector in investment-grade bond indices.

According to the latest report from JP Morgan, AI-related companies currently account for 14% of the investment-grade bond index, with total liabilities reaching USD 1.2 trillion. As the demand for AI infrastructure construction surges, related companies are heavily borrowing to support their high annual capital expenditure needs.

Oracle Model Triggers Debt Arms Race

Oracle has broken the previous model of AI infrastructure construction, which mainly relied on cash flow for funding. The company is willing to take on leverage of up to hundreds of billions of dollars to capture market share, forcing giants like Amazon, Microsoft, and Google to follow suit, further driving up the debt levels across the industry.

Previously, OpenAI committed to paying Oracle USD 60 billion annually for unbuilt cloud computing facilities. This news propelled Oracle's stock price up by 25%, but at the same time, Oracle's borrowing also increased, with the company's debt-to-equity ratio soaring to 500%, far exceeding Amazon (50%), Microsoft (30%), and Meta and Google.

Michael Cembalest from JP Morgan pointed out that the USD 60 billion annual expenditure promised by Oracle cannot be covered by cash flow and must be financed through equity or debt. This model is transforming the AI industry's originally disciplined, cash flow-financed competition into a debt-driven arms race.

Concerns Over Risk of Bubble Burst

Research from Bain & Company indicates that the construction of data centers needed to meet AI demand requires approximately USD 500 billion in capital investment annually. Bain's analysis of the sustainable capital expenditure-to-revenue ratio for cloud service providers shows that USD 500 billion in annual capital expenditure corresponds to USD 2 trillion in annual revenue.

Even if companies redirect all local IT budgets to cloud services and reinvest the funds saved from applying AI in sales, marketing, customer support, and R&D (estimated to be about 20% of these budgets) into new data center capital expenditures, they will still fall short by USD 800 billion compared to the revenue needed to cover all investments.

Analysts warn that once the AI paradigm shifts, whether due to the market suddenly demanding actual returns from AI investments or technological disruption, the bursting of the AI credit bubble could have more severe economic consequences than a stock market crash. If these debts, secured by future cash flows, default, it will impact the entire economic system