J.P. Morgan Asset Management warns: AI concentration risk is spreading to the bond market. Investors need to be vigilant about "crowded trades."

Gabriela Santos, a strategist at J.P. Morgan Asset Management, warned that the concentration risk of "artificial intelligence factors" has spread from the equity markets to the fixed income space. While the "supercycle" narrative still holds, the necessity for investors to act cautiously is rising sharply. In an interview on Tuesday, Santos stated, "You can be extremely bullish on everything related to AI, but you still need to think very carefully about portfolio construction." As the Chief Market Strategist for the Americas at J.P. Morgan Asset Management, Santos pointed out that the pullback in tech stocks in July highlighted the risk of crowded positions. During that month, the Philadelphia Semiconductor Index plummeted by 21%, marking its largest drop since 2008; at the same time, the Korea Composite Stock Price Index (Kospi) tumbled by 22%, with semiconductor manufacturers Samsung Electronics and SK Hynix accounting for about half of that benchmark's weight. Santos noted that this extreme volatility underscores the importance of position size control, leverage use, and diversifying investments outside of the AI sector. "This is where things get complicated. Because you can no longer just think about traditional factors, sectors, or regions, or even just asset classes, because the reach of AI is now everywhere," she added. She mentioned that U.S. Treasuries, gold, and core real estate are among the few areas that can provide different sources of returns. Santos pointed out that the expansion of AI infrastructureshe estimates capital expenditure in public and private markets will reach $5.5 trillionis "very unique," as it has already been reflected in corporate profits. She also indicated that growth rates will eventually slow, and even though AI remains a dominant investment theme, diversifying now is a wise move. Goldman Sachs strategists estimated in June that spending on AI data centers is expected to exceed $900 billion by 2026, with predictions for 2027 reaching as high as $1.4 trillion, highlighting the scale of the cycle described by Santos. Santos's warning about concentration extends to bond portfolios. She stated that the issuance of investment-grade bonds has set a historic high for the fourth consecutive month, with companies including Alphabet issuing 100-year bonds. Furthermore, multi-asset investors are now facing risks related to the expansion of AI infrastructure on both the equity and fixed income fronts. She noted that the debt issued by hyperscale cloud service providers should be studied individually rather than viewed as a whole, and she referenced the growing complexity of special purpose vehicles (SPVs) backed by data center leases.