Legendary investor Howard Marks, co-founder of Oaktree Capital Management and the prescient voice who warned about the dot-com bubble just three months before its March 2000 peak, has officially declared himself on “bubble watch.” In a recent memo to clients, Marks outlined several warning signs suggesting the stock market is approaching bubble territory, with artificial intelligence serving as the revolutionary technology fueling potential excess.
Marks drew parallels between today’s AI-driven market enthusiasm and previous bubbles he’s witnessed throughout his career. “The bubbles I’ve lived through have all involved innovations,” he noted, citing disc drive companies in the 1980s, internet stocks in the 1990s, and subprime mortgage-backed securities in the 2000s. AI now occupies that same position as the transformative technology driving investor excitement and potentially irrational exuberance.
However, Marks identified a crucial missing ingredient that prevents him from declaring a full-blown bubble: the widespread willingness of investors to buy stocks at any price. During the dot-com era, internet stocks would go public at already elevated valuations and triple on their first day of trading. “I don’t hear people saying, ’there’s no price too high,’” Marks observed, suggesting that while enthusiasm exists, it hasn’t reached the fever pitch characteristic of true bubbles.
The veteran investor acknowledged that market valuations are elevated but “not insane.” The S&P 500 currently trades at a forward price-to-earnings ratio of nearly 22x, above historical averages but not at extreme levels. “The markets, while high-priced and perhaps frothy, don’t seem nutty to me,” he stated.
Marks highlighted several cautionary signals worth monitoring: elevated investor optimism persisting since late 2022 (coinciding with ChatGPT’s introduction), above-average valuations, and the presumption that mega-cap tech companies will indefinitely lead markets higher. He warned that innovations leave investors without historical precedent, enabling “this time is different” thinking that often fuels bubbles.
To illustrate the danger of assuming today’s leaders will remain dominant, Marks noted that only six of the top 20 publicly traded US companies from early 2000 remain in that elite group today. “In bubbles, investors treat the leading companies — and pay for their stocks — as though the firms are sure to remain leaders for decades. Some do and some don’t, but change seems to be more the rule than persistence,” he explained.
Key Quotes
The bubbles I’ve lived through have all involved innovations
Howard Marks explained how artificial intelligence fits the pattern of previous market bubbles, comparing it to disc drive companies in the 1980s, internet stocks in the 1990s, and subprime mortgage-backed securities in the 2000s. This establishes AI as the current revolutionary technology driving potential market excess.
I don’t hear people saying, ’there’s no price too high'
Marks identified the key missing ingredient that would signal a full-blown bubble. This observation suggests that while AI enthusiasm is high, investors haven’t yet reached the irrational exuberance levels seen during the dot-com era when stocks would triple on their first day of trading.
The markets, while high-priced and perhaps frothy, don’t seem nutty to me
Despite being on “bubble watch,” Marks clarified that current valuations, while elevated with the S&P 500 at nearly 22x forward earnings, haven’t reached insane levels. This nuanced view suggests caution without panic regarding AI-driven market gains.
In bubbles, investors treat the leading companies — and pay for their stocks — as though the firms are sure to remain leaders for decades. Some do and some don’t, but change seems to be more the rule than persistence
Marks warned against assuming today’s AI leaders will maintain dominance indefinitely, noting that only six of the top 20 companies from 2000 remain in that elite group today. This has direct implications for how investors should value current AI market leaders.
Our Take
Howard Marks’ analysis provides a sophisticated framework for understanding AI’s role in current market dynamics. His “bubble watch” status is particularly noteworthy because he’s not declaring a bubble outright—he’s identifying the conditions that could lead to one. The AI parallel to previous revolutionary technologies is instructive: each innovation created genuine value while also enabling speculative excess. What’s fascinating is Marks’ observation about the missing ingredient—that final capitulation where investors abandon valuation discipline entirely. This suggests we may be in a middle phase where AI enthusiasm is justified but could tip into irrationality. The timing coincidence between elevated optimism and ChatGPT’s late 2022 launch underscores AI’s central role in current market psychology. For the AI industry, this serves as both validation of its transformative potential and a warning that market expectations may be running ahead of near-term realities. The historical reminder that corporate leadership changes dramatically over decades should temper assumptions about which AI companies will ultimately dominate.
Why This Matters
This analysis from Howard Marks carries significant weight for the AI industry and broader technology sector. Marks’ track record of identifying the dot-com bubble before it burst makes his current “bubble watch” status a serious warning signal for investors heavily positioned in AI stocks. The comparison between AI and previous revolutionary technologies highlights a pattern where transformative innovations often lead to overvaluation and eventual corrections.
For AI companies and their investors, this serves as a reminder that current market enthusiasm, while justified by AI’s potential, may be pricing in overly optimistic scenarios. The fact that only six of 2000’s top 20 companies remain in that tier today underscores the risk of assuming today’s AI leaders will maintain dominance indefinitely. This has implications for how businesses should approach AI investments and how workers should evaluate career moves to AI companies based on stock compensation. The timing is particularly relevant as AI deployment accelerates across industries, with valuations potentially disconnecting from near-term fundamentals. Marks’ observation that the final bubble ingredient—investors willing to pay any price—is still missing suggests there may be room for further gains, but also warns that such behavior could emerge as AI adoption continues.
Recommended Reading
For those interested in learning more about artificial intelligence, machine learning, and effective AI communication, here are some excellent resources: