Le coût de l'assurance contre les défaillances des entreprises d'IA augmente

Banks are scaling back lending as concerns mount over the growing disconnect between the massive debt burden fueling AI infrastructure and the economic returns it generates. Wall Street is increasingly wary of the risks associated with funding data centers and advanced computing hardware, which are central to the AI industry's expansion. According to research from Gartner, AI providers are expected to invest $7 trillion in infrastructure through 2030, yet they are forecasted to generate no more than $3.6 trillion in related revenue during the same period. This discrepancy raises alarms about the sustainability of current investment trends. A significant portion of the capital expenditure on AI infrastructure is financed through debt, which has led to increased scrutiny from banks and investors. In September 2026, global AI-related debt issuance dropped to $23 billion, a sharp decline from $113 billion recorded in June. As a result, financial institutions are becoming more cautious, with JPMorgan CEO Jamie Dimon indicating that his bank is tightening credit standards to mitigate exposure. CoreWeave, one of the leading neocloud providers, has been affected by these developments. Its stock and debt have faced a multi-day sell-off, with yields surpassing 15% and its stock price dropping by 15.5% before recovering slightly following comments from Nvidia CEO Jensen Huang. Similarly, Oracle's $160 billion debt is hovering near junk-rated status, prompting concerns about its ability to meet financial obligations tied to a data center project in New Mexico. The rising uncertainty is reflected in the increasing cost of insuring against defaults in the AI sector. For instance, the cost to protect $1 million of CoreWeave’s bonds against default has climbed from under $40,000 a year ago to $100,000 currently. Likewise, the cost of hedging against an Oracle default has risen by 12% in a week, with a $1 million policy now costing $28,000. These figures highlight the growing unease among investors regarding the financial health of major players in the AI industry. As the gap between AI spending and revenue widens, the future of the AI boom faces mounting challenges. With more than 80% of specialized AI cloud providers predicted to cease independent operations by 2030, the landscape for AI infrastructure is evolving rapidly. Companies are beginning to explore more efficient ways to utilize AI technologies, potentially altering the trajectory of the industry.