By Hamza Faisal

Howard Marks, the co-chairman of Oaktree Capital Management, says the artificial intelligence trade has lost its aura of invincibility — and he thinks that is a good thing. In an interview with Bloomberg Television on Monday, Marks said recent setbacks across AI-linked stocks have somewhat burst the bubble of infallibility that had surrounded the sector.

"I think people are now starting to say, 'Wait a minute, let's think more, let's be more skeptical,'" Marks said. "This is very healthy."

Marks, one of Wall Street's most closely watched investors, argued that enthusiasm for AI had turned the trade quite crowded, with money pouring in so fast that discipline slipped. "Investment standards have fallen while prices keep rising," he warned. "You have to act with caution."

A crowded trade under strain

The remarks land after a bruising stretch for AI stocks. One market tally put the decline in the combined value of the largest AI and chip companies at roughly $5 trillion since October 6, as investors question whether the trillions being poured into data centres and computing capacity will ever earn their keep.

Recent earnings reports from Alphabet and Amazon deepened the doubts. Both companies are spending enormous sums building AI infrastructure, raising questions about whether the core businesses generating the cash can keep funding the build-out. On Friday, Intel plunged nearly 29 percent after weak guidance, and the Nasdaq fell about 3 percent.

Marks sought to put the scale of the spending in perspective. "One trillion dollars is equivalent to earning one dollar a second for 31,000 years," he said. "No one can truly imagine what 31,000 years means, or understand the scale that implies."

The man who warned about dot-com

Marks carries unusual credibility on bubbles. In January 2000, he published a memo titled "bubble.com" warning about market euphoria weeks before the dot-com crash. In a December 2025 memo, he said he was not equipped to judge whether AI itself was a bubble, choosing instead to focus on who would capture the value — and whether the investment would pay off.

In late October, he argued the market had not yet reached its most dangerous moment, because investors were not oblivious to the risks. His test for a true bubble, he has said, is a public frenzy in which everyone not participating feels foolish.

Returns now under scrutiny

Marks is not alone in his caution. Jody Jonsson, vice chair of Capital Group, told Bloomberg Television separately that she was concerned about the gap between the scale of AI investment and the returns actually materialising. Bloomberg's reporting said investors are now questioning whether the industry's commercial gains can reward them adequately.

Marks has previously estimated that judging whether AI infrastructure spending pays off could take about a decade — a long wait for markets accustomed to quick returns.

For now, his message is not to flee AI, but to stop treating it as a sure thing. The next test arrives quickly: Apple reports earnings after the US close on Monday, with investors watching whether the world's most valuable company can stay above the fray.