AI Trade Sinks as Meta Bond Plan Exposes $680 Billion Credit Strain; Nasdaq Posts Worst Day in Months

U.S. AI stocks extended losses into the credit market, with the Nasdaq posting its worst single-day performance in months as investors reassess financing risk across the AI supply chain. U.S. high-yield risk premiums climbed to a 17-month high, and CCC-rated junk bond yields reached 15.85%, the highest since 2020. Meta (META) expects 2026 capital expenditures of $135 billion to $145 billion, with fourth-quarter spending of $49 billion to $59 billion. With free cash flow after dividends turning negative, management plans to raise hundreds of billions of dollars starting this fall—a pre-emptive move Barclays says reflects limited market capacity to absorb new AI debt. Goldman Sachs data show its AI basket has fallen about 55% from its June peak, erasing year-to-date gains, as hedge funds deleveraged in seven of eight sessions. Barclays estimates AI-related credit assets at roughly $680 billion, about 7.5% of the investment-grade index by market value but 13% of its duration-times-spread exposure. Spread dispersion among AI issuers has widened to 65 basis points from 24, with Oracle (ORCL), Meta and data-center operators underperforming more resilient short-dated debt from Microsoft (MSFT), Google (GOOGL) and Amazon (AMZN).