He Called 3 Market Crashes. Now He Warns of an AI Stock Bubble

When the people who make their living taking investor money start turning it away, it’s worth stopping to ask why. Two of China’s most respected hedge fund managers have done exactly that — and their reasoning points squarely at what they see as a dangerous AI stock bubble forming across Chinese and global equities.

Key takeaways

• Yang Dong of Wealspring Asset Management halted new fund subscriptions on November 1, 2025.
• Chen Guangming‘s Foresight Fund froze new subscriptions for its onshore hedge fund around the same date.
• Both managers believe AI-driven stock valuations have decoupled from fundamentals, forming what they call a super bubble.
• The Shanghai Composite Index hit a 10-year high in October 2025, fueled by global AI enthusiasm and improved US-China relations.
• Yang Dong previously called China’s market bubbles in 2007, 2015, and 2021 — each time correctly.

Two Top Chinese Hedge Fund Managers Shut the Door on New Capital

The move itself is almost unprecedented in its directness. Yang Dong, founder of Wealspring Asset Management, stopped accepting new fund subscriptions on November 1, 2025. His reasoning was not wrapped in vague macro caution — he pointed directly at AI-related equities and their increasingly untethered valuations.

Almost simultaneously, Chen Guangming‘s Foresight Fund followed with its own freeze on new subscriptions for its onshore hedge fund. Two independently operated funds, run by two of China’s most closely watched investors, arriving at the same conclusion at the same time. That kind of convergence isn’t coincidence — it’s a signal.

Their shared concern: prices have sprinted far ahead of the earnings reality that is supposed to justify them. Deploying fresh capital into that environment, in their view, means exposing new investors to losses that are not a matter of if but when.

The Market Context Behind the Warning

To understand the weight of this call, it helps to know what the market looked like in the weeks before they acted. The Shanghai Composite Index reached a 10-year high in October 2025 — a remarkable milestone driven by two powerful forces converging at once: surging global enthusiasm for artificial intelligence and a meaningful thaw in US-China diplomatic and trade relations.

Those tailwinds were real. Neither Yang Dong nor Chen Guangming disputed that. But real tailwinds and rational valuations are two different things. What they flagged was the gap between the excitement surrounding AI and the actual, verifiable earnings potential of the companies riding that wave. When sentiment outruns fundamentals by a wide enough margin, the eventual correction tends to be severe.

A Pattern Investors Have Seen Before

The dynamics are familiar to anyone who watched the dot-com era unfold. Tom Essaye, founder of Sevens Report Research, made the parallel explicit in a recent note, warning that cheap valuations on high-flying AI stocks — counterintuitively — may not be a buying signal at all. Instead, they could reflect investor skepticism that the earnings potential will ever materialize.

Essaye pointed to stocks like Nvidia, trading at 21x forward earnings, and Micron Technology, at just 10x forward earnings despite a 770% gain over 12 months, as examples of the distorted picture. Broadcom sits at 24x, while SanDisk has surged an extraordinary 4,490% over the past year at only 14x forward earnings. In a normal growth cycle, these stocks would command premium multiples. The fact that they don’t suggests markets are already pricing in doubt.

“This is exactly how the dot-com bubble burst,” Essaye wrote, drawing a direct line between today’s AI infrastructure build-out and the broadband overexpansion of the late 1990s. If major tech firms begin canceling data center projects because returns fall short of expectations, the knock-on effect across chip makers, memory suppliers, and networking companies could be swift and severe.

Why Yang Dong’s Warning Carries Particular Weight

Track record matters in this business, and Yang Dong has one that is genuinely hard to dismiss. He correctly identified the bubble that preceded China’s 2007 stock market crash. He flagged the 2015 Chinese equity bubble before markets shed roughly a third of their value in a matter of weeks. And he called the 2021 renewable-energy stock correction before that sector gave back its pandemic-era gains.

Three for three is not luck. It suggests a disciplined framework for identifying when price levels have structurally disconnected from underlying value — and the current AI stock bubble warning sits in that same tradition.

The subscription freeze itself is also more meaningful than a public statement would be. Managers can say almost anything in a market commentary. Turning away fee-paying clients costs real money. It’s the kind of action that reveals genuine conviction rather than performative caution.

Chen Guangming’s Parallel Decision Adds Institutional Weight

What makes the November 2025 moment particularly notable is that Chen Guangming, operating entirely independently at Foresight Fund, reached the same conclusion. Both managers cited difficulty finding investments that offer adequate risk-adjusted returns at current price levels. When two different institutions, with different portfolios and different investment processes, land on the same answer simultaneously, it speaks to something broader in the market structure rather than individual manager style.

What the Subscription Freeze Actually Means for Investors

The mechanism of protection here is straightforward but important. By closing to new subscriptions, Yang Dong and Chen Guangming are effectively saying: we do not want to deploy your capital at these prices. Existing investors who entered at lower levels are not forced out. But new investors who would buy in at the current peak are shielded from what both managers believe is a near-certain repricing.

It’s a quieter, more disciplined response than shorting the market or making dramatic public declarations. And in some ways, that restraint makes it more credible. They are not positioning to profit from a crash — they are simply declining to participate in what they consider an unsustainable valuation environment.

One notable detail: neither manager extended their concern to cryptocurrency or digital tokens. The AI stock bubble conversation, for China’s most prominent institutional fund managers, remains entirely a traditional equities story. That distinction matters for anyone trying to map the risk across asset classes.

The Dot-Com Parallel and Its Limits

The comparison to the dot-com era is powerful, but it requires some care. In the late 1990s, internet adoption was real and ultimately transformative — the mistake was in the speed of monetization expectations, not the technology itself. AI may follow a similar arc: genuine, lasting disruption, but a path to earnings that proves far slower and bumpier than the stock prices of 2025 implied.

Essaye’s framing captures this well. The issue isn’t whether AI matters — it clearly does. The issue is whether the scale and pace of data center construction, chip orders, and infrastructure investment can be justified by near-term revenue. Oracle’s roughly 25% share price decline since June 1, 2026, after heavy AI-related capital expenditure, offers a live example of what happens when the market starts questioning that equation.

For investors watching the Shanghai Composite Index and global AI-linked equities, the actions of Yang Dong and Chen Guangming represent something more valuable than a forecast. They represent a price. Two of China’s sharpest market minds have decided, with real capital on the line, that the current level is too high to enter. History suggests that kind of judgment is worth taking seriously — even if the timing of any correction remains, as always, the one thing no model can pin down.

FAQ

Why did Yang Dong and Chen Guangming stop accepting new fund subscriptions?

They believe AI stock valuations have inflated into a super bubble and want to protect investors from buying at unsustainable prices. Both managers cited difficulty identifying investments that offer adequate risk-adjusted returns at current market levels.

What market conditions contributed to the AI stock bubble warning?

The Shanghai Composite Index reached a 10-year high in October 2025, fueled by global AI enthusiasm and improved US-China relations. Those tailwinds pushed valuations in AI-connected equities significantly beyond what underlying fundamentals can support, according to both managers.

Have Yang Dong and Chen Guangming previously predicted market bubbles?

Yes. Yang Dong notably called China’s market bubbles ahead of the 2007 crash, the 2015 equity selloff, and the 2021 renewable-energy sector correction. His documented track record across multiple market cycles adds considerable weight to the current warning.

Does their warning include cryptocurrency or digital tokens?

No. Both managers focused exclusively on traditional equities and did not mention crypto assets or digital tokens. Their AI bubble concerns remain entirely within the stock market context.

Article produced with the assistance of artificial intelligence and reviewed by the editorial team.