Howard Marks: whether AI pays off is mostly a question of belief

what happened

Oaktree Capital co-founder Howard Marks argues that artificial intelligence is the most consequential and least predictable force investors have faced in a generation, according to a memo reported by Seeking Alpha. In it, he explains why top-down forecasting struggles here: an investor must get several linked variables right at the same time, and even a 60% hit rate on each one leaves roughly a 13% chance of success overall. Marks also cites J.P. Morgan data showing 42 AI-related companies generated about two-thirds of the S&P 500's gains since January 2024.

He flags debt as a second concern. Citing Goldman Sachs data, he notes that AI-related bond issuance has approached $600 billion globally this year, with about $400 billion going into U.S. investment-grade markets. He warns that hyperscalers willing to pay any price for funding could crowd out other borrowers, and that circular financing among chipmakers, frontier labs and data-center builders raises creditworthiness questions. Whether AI returns justify the trillions in capital spending, he wrote, is largely a matter of faith.

why it matters

For markets, the memo points to concentration risk: a handful of AI names drive broad index performance, so a shift in sentiment could hit the wider market. Marks also suggests data-center infrastructure is not an easy bet, with risks in power supply, cooling, regulation and chip depreciation. The tone is cautious rather than a call to action, and it is one investor's commentary, not a forecast.

market impact

AI stocks and the S&P 500: high concentration and debt load raise risks for the broader market

instruments

on polymarket