JP Morgan warns: up to $150 billion in leveraged loans in US CLOs facing AI risk

JPMorgan Chase said that between $400 billion and $1.5 trillion of leveraged loans packaged into U.S. collateralized loan obligations (CLOs) could be impacted by the artificial intelligence (AI) boom. The Wall Street bank pointed out that this is because these loans are concentrated in industries with the highest correlation to AI risk. This estimate was released during a review of the SFVegas 2026 conference, where the impact of software on corporate CLOs became the "hottest topic of the day". CLOs provide investors with exposure to floating rate debt, not fixed rate corporate bonds. The way they operate is by packaging leveraged loans into products with different levels of risk and return, similar to bonds, and then selling them to investors. Recently, CLO managers have been screening their portfolios to determine which loans are most vulnerable to AI influence, especially after the significant sell-off of software-related loans following the release of the powerful Claude Chat Siasun Robot & Automation by Anthropic PBC. In a report released on Thursday by JPMorgan Chase strategist Rishad Ahluwalia, it was written: "AI doomsday scenario? That seems a bit exaggerated. While focusing on the software industry is reasonable, we have told investors that we believe it is more important (though currently difficult to quantify) to consider the broader impact of AI disruption on CLO credit risk." To arrive at the estimated range of $400 billion to $1.5 trillion, the strategist used a simplified screening method based on market prices and rating information to assess AI credit risk in CLOs. However, they also acknowledged that this method still needs further refinement and improvement, and pointed to the healthcare industry as an example where it is difficult to provide investors with clearer judgments due to proprietary data issues and complex regulatory environments. The strategists also emphasized concerns about loan refinancing risk, noting that around $510 billion of software-related debt rated B- or lower will mature in 2028, and another $500 billion will mature in 2029. They wrote: "The significant exposure to the software industry in private credit suggests limited ability of the private market to refinance syndicated loan assets, which is unlike the more common scenario of transferring transactions from the public market to the private market in the past." The strategists stated that attendees at the conference also expressed concerns that if the labor market weakens or if anxiety around AI triggers more widespread sell-offs, it could bring about price risk. They said: "To be fair, our economists expect the diffusion of AI in the economy to be more gradual. However, the financial market's leveraging of AI also carries the risk of an unpleasant reset of expectations, and our cautious outlook for CLOs in 2026 reflects this theme."