Two of China’s most prominent hedge fund managers have issued stark warnings to their clients, declaring that the global frenzy surrounding artificial intelligence stocks has inflated into an unsustainable “super bubble.” The cautionary letters, which circulated widely in Chinese financial circles this week, argue that the massive capital expenditures on AI infrastructure are not being matched by revenue generation, and that a severe market correction is imminent.
The warnings from Wealspring Asset and Shanghai Banxia Investment Management Center represent a sharp, contrarian view from within China’s financial establishment, pushing back against the relentless optimism that has driven tech valuations to historic highs globally.
The ‘Collapse Point’ Approaches
Wealspring Asset, which manages over $1.4 billion and is led by Yang Dong, a manager famous in China for accurately calling the top of the 2007 stock market bubble, delivered a blunt assessment of the current AI landscape. In an investor letter reported by Bloomberg, Wealspring stated that AI stocks globally have entered “super bubble” territory and warned that “the collapse point may not be far away.”
Wealspring’s analysis drew a direct parallel between the current global rush into AI infrastructure stocks and the “brainless buying” that characterized China’s domestic stock market bubble in 2015. The fund predicted that the hottest Chinese AI infrastructure stocks, which have surged on the back of the global trend, could eventually crash by more than 80% when the bubble bursts.
Prior to these warnings, the CSI Artificial Intelligence Index had surged by more than 35% year to date, reflecting deep domestic enthusiasm for the sector.
The Anthropic Trigger
Shanghai Banxia Investment Management Center, managed by the highly influential Li Bei, went even further, arguing that the catalyst for the bubble’s deflation is already visible. In its client communication, Banxia pointed to pressure on Anthropic’s revenue growth as the specific trigger.
Banxia argued that Anthropic’s annualized revenue run-rate, a metric closely watched by AI bulls to justify massive valuations, is falling short of market expectations. The hedge fund predicted that large technology companies will soon recoil at the escalating costs of AI tokens, while intense competition will erode the market share of current leaders.
The core of Banxia’s thesis is a fundamental mismatch between infrastructure spending and application revenue. If the billions of dollars being poured into AI chips, data centers, and model training do not result in commensurate, sustainable revenue at the software and application layer, the current valuations of infrastructure providers cannot hold.
Market Jitters
The circulation of these hedge fund letters coincided with significant volatility in global semiconductor and tech stocks. On June 23, the Nasdaq Composite fell 2.2%, while South Korea’s tech-heavy KOSPI index plummeted nearly 10%, triggering a circuit breaker. US memory chip giant Micron Technology also saw its shares plunge more than 13%.
While it is difficult to attribute global market movements solely to the warnings of two Chinese hedge funds, their analysis taps into a growing undercurrent of anxiety among global investors regarding the sustainability of the AI trade.
The Capex vs. Revenue Debate
The warnings from Wealspring and Banxia are the latest salvo in a growing global debate about whether the extraordinary capital expenditures being poured into AI infrastructure can ever be recouped through software and service revenues.
The scale of the spending is staggering. Microsoft, Google, Amazon, and Meta have collectively committed to spending hundreds of billions of dollars on AI data centers, chips, and model training in 2025 and 2026. Nvidia, the primary beneficiary of this spending, has seen its market capitalization swell to over $3 trillion on the back of insatiable demand for its H100 and Blackwell GPU clusters.
Bull-case analysts argue that this infrastructure spending is analogous to the build-out of the internet in the late 1990s and early 2000s. While the initial wave of investment was followed by a painful correction, the underlying infrastructure ultimately enabled the creation of trillions of dollars in economic value over the following two decades. They contend that AI will follow a similar trajectory, with the current infrastructure serving as the foundation for a productivity revolution that will eventually justify the investment.
Bear-case analysts, including Wealspring and Banxia, counter that the comparison to the internet is flawed. The internet created genuinely new markets and business models. AI, they argue, is primarily an efficiency tool that compresses existing workflows rather than creating new ones. If AI primarily cannibalizes existing software markets rather than expanding them, the total addressable market may be far smaller than the bulls assume.
Context and Caveats
It is important to contextualize these warnings. As noted by industry observers, Banxia’s founder Li Bei issued this analysis during a period of significant pressure for her own firm. Banxia’s flagship fund reportedly suffered a 15.03% crash in the week of June 15 to 18, marking its worst single-week decline since inception. Consequently, the bubble warning must be read as analysis generated from inside a painful drawdown, rather than a detached observation from the sidelines.
Furthermore, the specific data Banxia cited regarding Anthropic’s revenue run rate is difficult to independently verify, as Anthropic remains a private company and its detailed financials are not publicly audited.
Nevertheless, the core argument presented by both Wealspring and Banxia, that the math between AI infrastructure costs and application revenues is becoming increasingly strained, is a testable hypothesis that is gaining traction among skeptical investors worldwide. As the AI industry matures, the market’s focus is inevitably shifting from the promise of future capabilities to the reality of current cash flows. If the revenue fails to materialize, the “super bubble” predicted by these Chinese funds may indeed burst.