US Stocks in Dot-Com-Style Bubble as AI Faces a Reality Check
Key Takeaways
- •Howard Marks of Oaktree Capital said AI companies are difficult to value and compared current US stock conditions to the late-1990s internet-stock mania that collapsed when the bubble burst in 2000.
- •South Korea's Kospi fell on Oct. 15, 2026 to its lowest closing level since April 7 as questions intensified over whether the debt-fueled AI boom will pay off, with losses exceeding those in other major markets.
- •South Korean retail investors directed nearly two-thirds of their net US stock purchases into a triple-leveraged semiconductor fund, leaning heavily on leveraged ETFs that magnify both gains and losses.
- •South Korea's National Pension Service saw its US stock portfolio climb to $155.1 billion as of the end of June, up 14.8% from the end of last year, driven by Big Tech and semiconductor holdings, while its China holdings surged 40%.
- •Audax Private Equity said private equity investors must look beyond traditional buy-and-build strategies of acquiring a platform company and expanding it through add-on acquisitions.

Artificial intelligence could become the most transformative force in the global economy, but investors face a significant challenge in determining what AI companies are actually worth, according to Howard Marks, co-founder and co-chairman of Oaktree Capital Management.
Marks's views were featured in KED Global's CIO Roundtable discussion, which ran under the headline "US stocks in dot-com-style bubble; AI faces reality check." The comparison recalls the late-1990s internet-stock mania, when investors bid up young web companies before valuations collapsed after the bubble burst in 2000. In his view, AI is hard to value, and he noted that interest rates remain moderate by historical standards — a detail relevant to valuation math, since share prices rest on discounting future earnings back to the present.
Audax Private Equity, a middle-market private equity firm, meanwhile, said private equity investors must look beyond traditional buy-and-build strategies, the industry term for acquiring a platform company and expanding it through add-on acquisitions.
Related Coverage
KED Global's coverage has tracked the AI boom's rise and the doubts now shadowing it.
South Korean investors are pouring money into US stocks, with nearly two-thirds of their net purchases going into a triple-leveraged semiconductor fund as they chase a rebound in American technology shares while Asian markets languish. Their buying has leaned heavily toward risky leveraged exchange-traded funds (ETFs), which use derivatives and borrowing to multiply an index's daily moves, magnifying both gains and losses.
South Korea's own stock market rode the AI boom to become one of the world's standout performers last year and through the first half of 2026, driven by Samsung Electronics Co. and SK Hynix Inc. That dependence, however, is becoming a liability as doubts mount over the sustainability of the AI-driven rally.
Those doubts turned acute on Oct. 15, 2026, when the Kospi, South Korea's benchmark stock index, fell to its lowest closing level since April 7. Renewed questions about whether the debt-fueled artificial intelligence boom will pay off sent South Korean stocks into a tailspin, with losses far exceeding those in other major markets, prompting warnings from analysts.
Institutional flows tell a similar story. South Korea's National Pension Service (NPS), one of the world's largest pension funds, saw the value of its US stock portfolio climb to $155.1 billion as of the end of June, up 14.8% from the end of last year, with the gains driven by Big Tech and semiconductor holdings, while its China holdings surged 40%.
Retail investors in South Korea, for their part, kicked off the year with a US equities-buying surge, snapping up more than $7.3 billion of shares in January as enthusiasm for Big Tech heavyweights such as Alphabet Inc. and Tesla Inc. eclipsed concerns over currency swings and government policy.
Source: KED Global