China’s leading AI-focused hedge funds are beginning to cash in on some of their biggest gains, reigniting debate over whether the AI rally is nearing its peak. Among them is Shanghai Everlead Capital, whose flagship fund has surged roughly 164% this year and has recently reduced exposure to several of its strongest-performing AI holdings.
The move does not necessarily signal expectations of an imminent market collapse. Instead, exclusive market-layer data indicates capital is rotating away from the most crowded AI trades, with investors increasingly focused on whether technology companies can sustain AI spending through 2027.
China’s Top AI Funds Take Profits
Everlead Capital recently trimmed positions in optical networking and chip-packaging companies—two areas closely tied to AI computing infrastructure.
According to the fund, valuations had climbed rapidly after a powerful rally. Companies such as Zhongji Innolight and Yangtze Optical Fibre posted extraordinary gains, reflecting investor enthusiasm for AI infrastructure. After such sharp appreciation, locking in profits became a logical portfolio move.
Another major player, Hunjin Capital, has adopted a similar strategy. The firm reduced exposure to heavily owned AI hardware names, including memory-chip companies that it believes could face weaker pricing power ahead. Capital has instead been redirected toward more attractively valued traditional sectors.
Hunjin also estimates that the current AI hardware investment cycle is about 60% complete, roughly twice the level it projected earlier this year.
While only two funds have publicly disclosed these portfolio adjustments, the broader question is whether the wider market is following the same pattern.
Market Data Points to Rotation, Not Retreat
Recent market performance suggests that investors are indeed rotating within the AI sector rather than abandoning it altogether.
AI compute stocks—including semiconductor and hardware companies—have delivered gains of roughly 62% over the broader rally but declined around 13% during the past month. Infrastructure and power-related companies also advanced earlier in the cycle before losing momentum.
Software and AI application companies have experienced the opposite trend. After underperforming for much of the year, the group has recently rebounded by approximately 5% as fresh capital flowed into the segment.
Rather than indicating the end of the AI trade, these shifts resemble a typical late-stage sector rotation, with leadership moving from infrastructure toward software.
Power Companies Have Become Part of the AI Trade.
Electricity providers and power infrastructure businesses were once driven primarily by interest rates and utility fundamentals. That dynamic has changed as AI data centers consume increasing amounts of electricity.
With global data-center power demand expected to roughly double by the end of the decade, energy companies have become closely linked to AI investment trends.
A proprietary indicator measuring the 30-day correlation between power stocks and AI compute companies has climbed to approximately 0.74, compared with near-neutral levels earlier in the cycle. The stronger relationship suggests that investors increasingly view power infrastructure as an extension of the AI ecosystem.
That relationship also creates downside risk. When AI hardware stocks weaken, power-related shares often move lower as well, making future AI investment spending the key variable for both sectors.
Why 2027 AI Spending Could Determine the Outcome
The central issue remains AI capital expenditure, the billions of dollars major technology companies continue investing in chips, data centers, networking equipment, and electricity infrastructure.
As long as those investments continue expanding, recent selling by hedge funds could simply represent prudent profit-taking after exceptional gains.
Industry forecasts estimate that the world’s largest cloud and technology companies will spend more than $600 billion on AI infrastructure in 2026, representing annual growth of roughly 36%. Some projections suggest cumulative investment could exceed $1 trillion by 2027.
For now, those forecasts continue supporting the bullish case.
However, investors are increasingly watching for signs that spending could plateau beyond 2027.
One potential risk comes from intensifying competition. Chinese AI models are approaching the performance of leading U.S. systems while reportedly operating at dramatically lower costs. As cheaper models become more capable, they could reduce the economic return generated by massive infrastructure investments.
If profitability declines, technology companies may eventually slow AI capital spending. That would weaken demand for chips, networking equipment, and power infrastructure, potentially validating today’s profit-taking as an early warning of a larger correction.
China is about to pop the AI bubble.
Every industry China has entered has seen profit margins collapse.
Yet somehow people believe AI will be the exception.
1. Market share
Chinese AI models have accounted for more than 30% of US -originating token traffic on OpenRouter… pic.twitter.com/XHOPH8lqDR
— Lukas Ekwueme (@ekwufinance) July 15, 2026
Supporters of the AI investment boom argue that current valuations remain backed by genuine earnings growth rather than speculative excess. Skeptics believe mounting pricing pressure could eventually undermine those returns.
Ultimately, the trajectory of AI capital expenditure in 2027 may become the deciding factor. Continued spending would reinforce the view that recent selling reflects normal profit-taking. A sharp slowdown, however, would strengthen arguments that the AI boom had developed into a bubble.
