Shanghai Everlead Capital told investors in its Growth Strategy No. 3 Fund, which returned 164% this year through May 31, that it has trimmed positions in optical communications and advanced packaging companies, according to an investor letter seen by Bloomberg News.

Hunjin Capital's Yueyang G1 fund, which was up by a third in the first five months of the year, sold some of its AI stocks, according to its May investor letter. The firm, which manages more than 5 billion yuan ($739 million), said it had identified specific triggers that would justify a broader exit.

China's AI bulls aren't yet warning the bubble is about to burst, in contrast with some of their more skeptical rivals. But an examination of performance letters recently sent to investors in the country makes clear that even funds that have generated bumper returns from the AI trade are now starting to get nervous.

"You can take advantage of the frenzy, but never become the frenzy," wrote Dan Bin, a well-known fund manager at Shenzhen Oriental Harbor Investment Management Co., in the firm's May letter.

Dan told investors that 2027 may be the year when AI capital expenditure reaches an inflection point, finally slowing down after years of breakneck growth. His firm runs more than 10 billion yuan in assets.

Chinese funds join a cohort of investors concerned that this year's surge in AI-related stocks, including semiconductor and memory-chip companies, may have gone too far, too fast. The Philadelphia Semiconductor Index has soared almost 80% this year alone, powered by Micron Technology Inc. and Intel Corp.

The global rally in AI stocks has been a dividing line in Chinese equity funds' performance this year. Many of those betting on AI have posted big gains, while funds that avoided the sector are largely trailing the market.

Trigger Points

Hunjin Capital told investors that they should be on the lookout for a social backlash against AI, a plateau in model improvement and a loosening supply chain for the technology. It said storage makers could see their huge profit margins squeezed as AI firms find more alternatives.