The intense battle between investment conviction, generational understanding, and human nature is reaching a fever pitch within public and private fund institutions, as the shadow of the former "king" Kweichow Moutai fades on the A-share market, making way for new tech elites like Cambricon Technologies, Yuanjie Semiconductor, and Lianxun Instrument to take turns at the top.
During the fourth week of June, institutional investors in the unpredictable ocean of A-share capital are experiencing extreme emotions of joy and sorrow, and a deep divide.
In a conference room at a private equity firm in Shanghai's Lujiazui, the atmosphere is so heated it feels almost frozen. On the screen, the share price of Suzhou Lianxun Instrument is piercing through the 2,500 yuan per share barrier in a manner that seems to "defy gravity." This optical communication equipment manufacturer, listed for only two months, has sparked a market value frenzy worth hundreds of billions of yuan in just a few days. It has become one of the case studies for internal discussions at this private equity firm.
Resonating with this is the collective surge of products under fund manager Jin Zicai of Caitong Fund Management Co., Ltd.. The premium rate of the Caitong Duoce Fuxin LOF once soared as high as 51%. Just over a year ago, Jin Zicai was a "disappointed" figure, with the performance of his managed funds dropping over 30%, placing him near the bottom of rankings. Today, relying on an extreme overweight position in AI computing power, he holds several "miracle funds" that have surged over fivefold in a year, being frantically pursued by millions in follow-on capital.
However, during the same period, several letters were being delivered to the inboxes of high-net-worth clients, their wording tinged with bitterness. Yang Dong, the helm of Ningquan Asset Management, hailed as the "conscience of the industry" for his precise exit before the bear market arrived, typed out a sincere "apology" to investors. Because he refused to participate in the AI infrastructure bubble with price-to-earnings ratios reaching hundreds of times, the net asset value of the products he manages saw a significant decline in June. Almost simultaneously, Li Bei of Banxia Investment, known as the "Financial Goddess," handed in a report card showing a single-week drawdown of 15%. Facing client inquiries and potential redemption pressure, she wrote a warning in her monthly report: "Even if you scold me, I still want to advise you to be extremely cautious about chasing AI."
On one end is the myth of wealth creation, blazing hot with gains exceeding fivefold in a year; on the other is a bitter adherence to principle, treading on thin ice, preferring to suffer net value setbacks rather than compromise.
The intense battle between investment conviction, generational understanding, and human nature is reaching a fever pitch within public and private fund institutions, as the shadow of the former "king" Kweichow Moutai fades on the A-share market, making way for new tech elites like Cambricon Technologies, Yuanjie Semiconductor, and Lianxun Instrument to take turns at the top.
Feverish Dance
If a bull market is the epitaph for pessimists, then an extreme structural market is the coronation for high-stakes gamblers.
In this tech stock-led market, Caitong Fund Management's fund manager Jin Zicai is one of the most dramatic protagonists.
Rewinding the timeline to a year ago, Jin Zicai was a "forlorn figure" fading at the bottom of performance charts. In June 2025, several products he managed fell over 30%, placing them at the bottom among similar funds. However, according to Wind data, as of June 24 this year, just one year later, the six products under his management that have been operating for a full year all posted returns exceeding 400% over the past year. Among them, Caitong Duoce Fuxin achieved a staggering return of over 535.20%, becoming an exceedingly rare "five-bagger" fund in the entire market.
The reversal from trough to peak for Jin Zicai can be described as "simple and brutal"—an extremely concentrated portfolio.
A deep dive into the periodic reports of his managed funds reveals that he has highly concentrated the fund's bets in the computing power and communications sectors. The top holdings of his seven managed products cover only 11 individual stocks, with the top ten holdings accounting for over 90% of the stock investment market value for several consecutive quarters. Stocks like Sunny Optical, Zhongji Innolight, and Dingtai High-Tech—superstar stocks that have multiplied several times over during the AI wave—are held in heavy positions by him using a "seven funds, one strategy" model.
When the trend is favorable, this strategy is like a war chariot loaded with all its ammunition, possessing nearly unmatched explosive elasticity.
The dazzling performance of the funds has ignited a frenzy of investor capital inflows.
On June 23, Caitong Fund Management announced the suspension of large-amount subscriptions, regular fixed-amount investments, and conversion transfers for four of its funds, including Caitong Value Momentum Mixed, Caitong Quality Selection Mixed, Caitong Integrated Circuit Industry Stock, and Caitong Growth Preference Mixed.
A few days earlier, on June 19, two funds, Caitong Masterpiece Selection One-Year Holding and Caitong Prosperity Selection One-Year Holding, set a purchase limit of 1,000 yuan per single account per fund. Adding the closed-end Caitong Duoce Fuxin LOF, all seven funds currently managed by Jin Zicai have implemented purchase restrictions. Besides limits, since May this year, Caitong Fund Management has continuously issued premium risk warnings for on-exchange funds like Caitong Fuxin LOF, even resorting to intraday trading halts to cool down the fervor.
Is it a legend, or a gamble? In the current A-share market, the two are often separated by only a thin line.
Apologies Issued
In stark contrast to Jin Zicai are the bitter perseverance and heavy sighs of some former investment veterans. As capital is siphoned into the tech sector, seasoned managers who adhere to traditional value investing and respect market cycles are perhaps experiencing the most severe crisis of confidence in their careers.
Yang Dong, the helm of Ningquan Asset Management, began his recent second-quarter letter to holders with a weighty "apology."
Yang Dong holds heavy positions in Chinese internet giants that possess vast data, computing resources, and ample cash flow, believing they are the ultimate winners in the AI marathon. However, these giants, whose market capitalizations are far inferior to their US counterparts, have shown a puzzling downward trend over the past year. He positioned in utility stocks that should have benefited from AI power demand, only to watch US power stocks double while domestic power stocks remained stagnant due to supply-demand dynamics.
Facing the AI infrastructure sector's surge, Yang Dong frankly stated in his letter that the company's performance this year was not up to par, apologizing to holders, and直言 the team "severely underestimated the scale of the bubble the AI infrastructure theme could form." "A large number of popular A-share stocks are highly likely to fall by 80% or even over 90% in the future; we dare not participate anymore." Yang Dong further stated that he would rather endure the largest drawdown since the product's inception, would rather "pick up cigarette butts" in real estate, building materials, and solar sectors, than cater to the market.
Similarly unyielding is Li Bei of Banxia Investment, the "Financial Goddess."
Under immense pressure from a single-week net value plunge exceeding 15% and a year-to-date decline of over 20%, Li Bei maintained her characteristic sharpness. Her heavily weighted sectors—energy, real estate, consumer goods, and building materials—have suffered in the extremely polarized market. In her letter to investors within the monthly report, she stated bluntly: "If investors have lost patience and no longer wish to see our analysis and outlook, I completely understand. If any investors wish to redeem their funds, hold cash, and wait for a clear uptrend to re-enter, I also respect that. But if investors want to use that money to chase AI, even if you scold me, I still want to advise you to be extremely cautious."
Li Bei's bearish stance does not stem from a rejection of new things. She believes that as a representative of the most downstream model companies, the annual recurring revenue growth rate of Anthropic has slowed significantly. It is highly likely to be significantly lower than the market's previous optimistic expectations by year-end, making a subsequent decline in capital expenditure expectations highly probable.
Clashing Views
Looking beyond the surface of stock prices, the unprecedented divergence among institutional investors regarding tech stocks today essentially stems from differing views on "what stage of the lifecycle the current AI industry is in." On the debate tables of these professional investors, the logics of bulls and bears are clashing fiercely.
The confidence of the bulls stems from the un-falsifiable industry explosion and overseas parallels.
"Embrace camp" representatives like Wang Guizhong, Director of Big Tech Research at Harvest Fund Management and fund manager of Harvest Science and Technology Innovation Fund, believe that the global AI industry officially moved towards a commercial closed loop in the first quarter of this year, with Agent + Coding sweeping the globe in the form of productivity tools. More directly, there is a strong parallel from overseas fundamentals. On June 24, US Eastern Time, memory giant Micron Technology delivered earnings that exceeded Wall Street expectations, with quarterly revenue surging 346%. Its CEO Sanjay Mehrotra pointed out that the data center business annual run rate has broken through the $100 billion mark, and High Bandwidth Memory has become the tightest segment after GPUs, with Micron's full-year 2026 HBM capacity already fully booked under fixed-price contracts.
Li Yue, founder of Hun瑾 Capital, stated bluntly in his monthly report not to apply the 1999 dot-com bubble template to this round, with the core difference being profit drivers.
Citing a leading optical module company as an example, he said its profit was only around 1 billion yuan in 2021, reaching 45 billion yuan this year, and the profit corresponding to customer orders already placed for next year is between 80-100 billion yuan. With profits up 100-fold, next year's P/E would only be around 15 times. Calling this a "bubble" is inappropriate.
The bears' counterattack stems from perceived "flaws" in the business model and the gravitational pull of mean reversion. Investors from this segment of institutions believe that undercurrents are already swirling beneath the surface prosperity.
Xia Junjie, founder and Chief Investment Officer of Ren桥 Asset Management, stated that those inside the circle see only flames, while those outside see only ice.
He believes that large models require perpetual massive capital expenditure, GPU depreciation cycles are only five years, and coupled with the variable cost of electricity behind token consumption, the ROI may simply not add up.
He cited two examples: Uber exhausted its annual token budget in the first four months, and Microsoft suspended its employees' free direct access to Claude. He believes that before "physical AI" is realized, the fields digital AI can affect have a limited total value, similar to the 2000 internet bubble, where the actual capacity and market value are severely mismatched at this stage.
Ridou Investment also admitted in internal communications that the AI frenzy is concentrated in the hardware field, with its development speed exceeding expectations, leading to short-term capital concentration. But in the long run, a healthy economy requires balanced development across industries; it's impossible for all resources to be permanently directed towards a single direction. Moreover, investment must see returns. If continuous high-intensity investment fails to achieve profitability, capital will eventually recede.
Even Butbin of Orient Harbor, who is in the bull camp, retains a dose of清醒. He worries about the impact of memory price increases on the profit margins of downstream Cloud Service Providers. The prosperity of AI applications极度 relies on a cliff-like drop in computing power costs, and rising hardware prices run counter to this ultimate requirement.
Bulls judge the AI trend will continue, while bears predict market repricing after the AI bubble bursts. Both sides have supporting logic, and both have potential flaws.
Why is the divergence so intense? In the view of several interviewed institutional investors, the A-share market is currently in a special period of a vacuum in macro narratives and intense micro-liquidity competition. When domestic consumption is weak and real estate is still struggling to find a bottom, capital urgently needs a sector with infinite imagination space and endorsement from overseas giants' performance to herd into. AI has become that outlet.
However, while technological innovation indeed develops exponentially, this does not mean capital market pricing can ignore physical laws and商业常识.
As Yang Dong said, "Every industry has its prosperous times... supply will eventually catch up, especially since supply growth driven by暴利 often exceeds expectations."
Comments