Key Takeaways
- • Two major Chinese hedge funds—Wealspring Asset ($1.4B AUM) and Shanghai Banxia Investment Management Center ($294M AUM)—have publicly declared the global AI equity trade a "super bubble" and are actively rotating out of core positions in Nvidia and US hyperscalers (Microsoft, Amazon, Alphabet, Meta).
- • Wealspring's founder Yang Dong carries significant credibility on bubble identification, having previously predicted the peak of the Chinese stock market in 2007, lending weight to current warnings about AI valuation excesses.
- • Shanghai Banxia cited slowing revenue growth at Anthropic as evidence that conditions for bubble collapse are already forming, suggesting deteriorating fundamentals beneath the surface of AI sector enthusiasm.
- • AI-focused hedge funds experienced significantly declining average excess returns in the first half of 2026, with July 2026 delivering steep sector drawdowns marked by unusual price volatility that forced some funds to cut positions or inject proprietary capital to avoid forced liquidations.
- • The rotation is not a complete exit from AI exposure but rather a strategic shift into the broader data center supply chain and hardware enablers positioned further from the spotlight of marquee hyperscaler stocks.
Summary
Two of China's most prominent hedge funds have issued urgent warnings to investors about unsustainable valuations in global AI equities, signaling a strategic retreat from the sector's flagship names. Wealspring Asset, managing approximately $1.4 billion in assets, characterized the current AI trade as a "super bubble" in a recent client letter, arguing that its collapse could materialize sooner than consensus expectations. Shanghai Banxia Investment Management Center, a smaller firm with roughly $294 million under management, reached a similar conclusion through fundamental analysis, pointing specifically to deceleration in Anthropic's revenue growth as evidence that the preconditions for a bubble burst have already begun to manifest.
Both funds have commenced systematic reductions in exposure to the stocks that have anchored the AI trade narrative: Nvidia and the four major US hyperscalers—Microsoft, Amazon, Alphabet, and Meta. Rather than abandoning AI exposure entirely, the funds are executing a nuanced repositioning into the broader ecosystem of data center enablers and hardware supply-chain companies that occupy less prominent positions in market discourse. This distinction reflects a view that while AI infrastructure demand may remain robust, the valuations of core hyperscaler and semiconductor stocks have become untethered from fundamentals.
The timing of these warnings carries particular weight given market dynamics in 2026. AI-focused hedge funds reported significantly compressed excess returns throughout the first half of the year, according to industry data. July 2026 intensified this pressure, delivering steep drawdowns across AI-exposed portfolios as sector prices exhibited unusual volatility and momentum reversals. Fund managers responded with divergent approaches: some funds executed outright cuts to AI positions, while others deployed proprietary capital to stabilize portfolios and avoid forced liquidations at depressed prices.
Wealspring's founder, Yang Dong, brings substantial credibility to the fund's bubble thesis. He is widely recognized in China for accurately predicting the peak of the Chinese stock market in 2007, establishing a track record of identifying inflection points in equity market cycles. This historical precedent lends institutional weight to current warnings about AI sector valuations, particularly for investors evaluating contrarian signals from experienced market participants.
Both funds have commenced systematic reductions in exposure to the stocks that have anchored the AI trade narrative: Nvidia and the four major US hyperscalers—Microsoft, Amazon, Alphabet, and Meta. Rather than abandoning AI exposure entirely, the funds are executing a nuanced repositioning into the broader ecosystem of data center enablers and hardware supply-chain companies that occupy less prominent positions in market discourse. This distinction reflects a view that while AI infrastructure demand may remain robust, the valuations of core hyperscaler and semiconductor stocks have become untethered from fundamentals.
The timing of these warnings carries particular weight given market dynamics in 2026. AI-focused hedge funds reported significantly compressed excess returns throughout the first half of the year, according to industry data. July 2026 intensified this pressure, delivering steep drawdowns across AI-exposed portfolios as sector prices exhibited unusual volatility and momentum reversals. Fund managers responded with divergent approaches: some funds executed outright cuts to AI positions, while others deployed proprietary capital to stabilize portfolios and avoid forced liquidations at depressed prices.
Wealspring's founder, Yang Dong, brings substantial credibility to the fund's bubble thesis. He is widely recognized in China for accurately predicting the peak of the Chinese stock market in 2007, establishing a track record of identifying inflection points in equity market cycles. This historical precedent lends institutional weight to current warnings about AI sector valuations, particularly for investors evaluating contrarian signals from experienced market participants.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.