• Howard Marks warned that the AI hype is fueling risky "moonshot" bets on startups with no profits.
• He says most AI newcomers will fail — and prefers tech giants that can absorb the shift.
• The Oaktree Capital cofounder added that new tech rarely guarantees profits for investors.

Legendary investor Howard Marks has seen enough manias to know when a bubble is brewing.

On the "We Study Billionaires" podcast last Saturday, the billionaire and cofounder of Oaktree Capital Management said he is alarmed by how many investors are piling into speculative startups in hopes of landing the next trillion-dollar winner.

"Do you want to have a novel entrepreneurial startup pure play which has no revenues and no profits today, but could be a moonshot if it works?" he asked.

"Or do you want to invest in a great tech company, which is already existing and making a lot of money where AI could be incremental but not life-changing? It's a choice."

Too many investors, he said, are treating AI like a lottery where the slim chance of a jackpot overshadows the far greater likelihood of failure.

"People say, 'Well, they have a low probability of success, but maybe a big payoff, so I should buy it.' That's what I call lottery-ticket mentality," he said.

A familiar bubble pattern

Marks has been sounding this warning for months.

In a 2023 conversation with financial historian Edward Chancellor on Oaktree's "Behind the Memo" series, he put it bluntly: "AI will change the world," he said, predicting that most of the companies that people are investing in today for AI purposes "will end up worthless."

On the "We Study Billionaires" podcast, Marks said he sees echoes of the dot-com era: a groundbreaking technology, sky-high expectations, and a growing assumption that early movers will inevitably dominate.

"Most bubbles are around something new," he said, pointing to the growth-stock boom of 1969, the subprime-mortgage mania of 2006, the dot-com bubble in 1999, the 1720 South Sea Bubble, and even the Tulip Craze of 1636 in the Netherlands.

Such moments, he added, arise because "the imagination is untrammelled, and it can go off in a flight of fancy."

In reality, he said, most early-stage companies don't survive the transition from promise to profitability.

"On the AI, I'm led to believe that you can make binary bets in companies that have nothing else going on, which will be sink-or-swim bets, or you can invest in pre-existing great tech companies, which will get moderate benefits from AI if they're successful."

Why Marks prefers tech giants over AI pure plays

Rather than gamble on moonshots, Marks said he'd rather own stocks in established tech titans that are already generating profits — companies that can integrate AI as an incremental advantage rather than a life-or-death proposition.

He contrasted those firms with the fragile AI startups drawing hype, saying that Big Tech is positioned to reap some benefits from AI, even if the technology rolls out more slowly or unevenly than hoped.

But a breakthrough in productivity doesn't automatically translate into investment gains.

"Change the world and investors making money are not the same thing."

Marks echoed a warning Warren Buffett issued at Berkshire Hathaway's 2000 annual meeting, when Buffett warned that the internet stock mania had detached valuations from underlying profit potential — a mistake he believes investors are repeating with AI.

"I think the same is true of AI," he said.