On July 27th, the A-share market exploded once more. ChangXin Memory Technologies officially debuted on the STAR Market, opening at 49.5 yuan, a staggering 471.59% surge from its 8.66 yuan issue price, catapulting its total market capitalization to 3.31 trillion yuan. This immediately surpassed ICBC to claim the top spot on the A-share market cap leaderboard. For the first time in STAR Market history, a company breached the 3 trillion yuan ceiling at the open, and for the first time ever, a tech stock ascended to the pinnacle of A-share market value. Within the first hour of trading, turnover exceeded 103.5 billion yuan, with the full-day closing turnover settling at 141.2 billion yuan. To put that figure in perspective, the previous A-share single-stock single-day turnover record was 90.038 billion yuan, set by East Money on October 9, 2024. ChangXin Memory has directly raised that record by over 50%. The 141.2 billion yuan is equivalent to turning over an entire year's GDP of many provinces through just one stock. The trading was so frenzied that many brokerage trading systems experienced glitches and order delays. One retail investor joked, "I wasn't excited about getting allocated shares, but the system freezing scared the sweat out of me." The stock peaked during the session at 55.03 yuan, representing a gain of up to 535.45% and pushing the total market cap past 3.6 trillion yuan. This figure surpassed Intel's market cap of 465.6 billion USD (approximately 3.15 trillion yuan) on the US market on July 24. The stock closed the day at 49 yuan, giving a total market cap of 3.28 trillion yuan. With 500 shares per lot, winning an allocation netted a profit of roughly 20,000 yuan. For many retail investors who hadn't won an IPO allocation in years, this felt like a windfall from heaven.

But amidst the joy of the lucky winners, a more poignant story circulated in the market.

Country Garden: The Most Regretted 'Sold-too-soon' in A-Share History

Let's rewind to 2021. Back then, Country Garden was still one of the top-selling developers in China, flush with cash, and its venture capital arm was actively seeking a second growth curve. That year, Country Garden's venture capital division invested 900 million yuan in ChangXin Memory as a Series B strategic investor, securing a 1.56% stake. At that time, ChangXin's overall valuation was only around 40 billion yuan. Looking back from today's vantage point, the foresight of this investment was textbook-perfect. For a real estate company to cross over and invest in semiconductors, accurately betting on China's only DRAM champion—this level of judgment surpassed the vast majority of professional investment institutions. Unfortunately, the story did not have a fairytale ending. The winter in the real estate industry hit harder and faster than anyone anticipated. After 2023, Country Garden's liquidity pressure surged dramatically. Delivering homes, repaying debts, and paying wages—everything required hard cash. Survival became the overriding priority. On December 27, 2024, Country Garden announced in a filing to the Hong Kong Stock Exchange that it had transferred its entire 1.56% stake in ChangXin Memory to the Hefei state-owned capital platform, Hefei Jianchang, for a total consideration of 2 billion yuan. A 900 million yuan entry, a 2 billion yuan exit—a book profit of 1.1 billion yuan. That sounds pretty good, right? But what if they had held until today? Based on the closing market cap of 3.28 trillion yuan, a 1.56% stake would be worth over 51.1 billion yuan. Country Garden missed out on nearly 50 billion yuan. What does 50 billion yuan mean? Country Garden's current total market cap on the Hong Kong stock exchange is just over 8 billion Hong Kong dollars. This missed fortune could buy six of today's Country Garden. Even more painful is the transfer price—approximately 2.22 yuan per share. This price was only one-quarter of the IPO price and just 4.5% of the opening price. Calling it a fire sale at a bargain price is no exaggeration. No one can really blame Country Garden. Given their predicament at the time, the 2 billion yuan in cash was a lifeline. The cruelest part of the capital market is never about being wrong, but about being right and not surviving long enough to cash in. It wasn't poor judgment; it was that cash flow couldn't sustain ambition. It wasn't a failure to see the future; it was a failure to endure the present.

While some are crying, others are laughing: Alibaba's paper profit exceeds 150 billion, and Hefei's state capital reaps massive rewards. Also an early investor, some fall before dawn, while others taste the sweetest fruit. Alibaba is the industrial investor with the highest shareholding percentage in this capital feast. Through its entities Alibaba Cloud (holding 3.85%) and Alibaba Network (holding 1.12%), Alibaba holds a combined 4.97% stake in ChangXin Memory, with a cumulative investment of approximately 7.6 billion yuan. In the Pre-IPO round in June 2025, Alibaba Cloud alone poured in 6.1 billion yuan. Based on the 3.28 trillion yuan market cap, Alibaba's stake is worth over 163 billion yuan, yielding a paper profit exceeding 155 billion yuan. But the brilliance of this investment goes far beyond financial returns. ChangXin is now a core DRAM supplier for Alibaba Cloud, offering server memory that is about 30% cheaper than imported equivalents. In the AI computing race, the stability and cost advantage of the upstream supply chain are even more valuable than the paper profit. But Alibaba is not the biggest winner. The true super-winner is the Hefei state-owned capital system. When ChangXin Memory was founded in 2016, DRAM technology had high barriers, required massive investment, and had long R&D cycles. Globally, only Samsung, SK Hynix, and Micron could play the game. The industry's prospects were highly uncertain; no one knew if China could crack this tough nut. It was against this backdrop that Hefei Industry Investment contributed approximately 14.4 billion yuan out of the 18 billion yuan budget for the first phase, shouldering 80% of the initial investment. This wasn't an investment; it was a high-stakes gamble. A provincial capital city in central China was betting its fiscal resources on a track monopolized by only three companies globally. Over a decade, the total investment reached approximately 24.8 billion yuan. Today, the Hefei state-owned capital system holds a combined stake of approximately 36.79% through platforms like Qinghui Jidian, ChangXin Jicheng, ChanTou YiHao, and Hefei Jianchang. Based on the closing price, this corresponds to a market value of over 1.2 trillion yuan, with a paper profit exceeding 1 trillion yuan. Ten years, 24.8 billion yuan turned into 1 trillion yuan. This is not something that can be summed up by the title "Best Venture Capital City"—it is the most stunning footnote in the history of Chinese industrial investment. Additionally, the National Integrated Circuit Industry Investment Fund Phase II holds an 8.73% stake, corresponding to a market value of approximately 288.9 billion yuan; the Anhui Provincial Investment holds a 7.91% stake, valued at around 216 billion yuan. Provincial and municipal state capital combined hold about 45.45%. Chairman Zhu Yiming indirectly holds approximately 1.59 billion shares through multiple platforms, with a market value of about 78.7 billion yuan. More notably, Zhu Yiming has pledged to distribute 760 million of his ChangXin Memory shares to employees for incentives over 10 calendar years, starting 3 years after the company's listing. At 49 yuan per share, these shares are worth over 37.2 billion yuan. This single IPO has created at least 2,370 millionaires. The company's two phases of employee stock ownership plans have awarded shares to a total of 6,760 person-times. The grant price for the 2023 plan was a mere 0.108 yuan per share—a 453-fold increase. This is the wealth-creating magic of the semiconductor industry, and the era's dividend from domestic substitution.

Is a 3 trillion yuan valuation expensive? Nomura sees 7.76 trillion. On the listing day, Nomura swiftly issued its inaugural coverage report, directly giving a target price of 116 yuan, corresponding to a total market cap of about 7.76 trillion yuan. What level is this target price? It's 1,239% higher than the issue price, implying the stock needs to more than double from its current price. Why is Nomura so bold? The core logic has three layers. First, the global DRAM supply landscape is undergoing structural change. The capital expenditure of the three major players—Samsung, SK Hynix, and Micron—is shifting massively towards high-end HBM storage, actively compressing new capacity for general-purpose DRAM. In the coming years, supply of general-purpose DRAM is unlikely to be loose. Second, ChangXin's market share growth could significantly exceed market expectations. ChangXin currently holds a 7.67% global share, firmly ranking fourth globally. Nomura believes its long-term market share could reach 25% to 30%. Once the growth trajectory accelerates, the potential is considerable. Third, there is a dual premium from domestic substitution and AI demand. As the only pure-play DRAM stock on the A-share market, its scarcity itself commands a premium. Nomura predicts that from 2026 to 2028, ChangXin's sales and net profit compound annual growth rates will reach 63% and 74% respectively. Using a projected 2028 EPS of 5.8 yuan and a 20x P/E ratio gives the 116 yuan target price. Of course, this is the most optimistic view. The consensus among foreign institutions is in the 3 to 5 trillion yuan range, while domestic brokerages hold a neutral view of 2 to 3 trillion yuan. The wide divergence in opinion highlights the immense potential for imagination around this stock.

Are the Koreans panicking? Behind the 950 Billion USD mega-deals. The day ChangXin Memory listed was also unsettling for the South Korean market. Just a few days prior, Kim Yong-beom, the head of the South Korean Presidential Office's Policy Office, announced a staggering figure in San Francisco: 950 billion USD. Samsung Electronics signed a business cooperation MOU with Broadcom for wafer foundry collaboration around a 200 billion USD scale of advanced memory semiconductor supply and AI chip manufacturing over the next five years. SK Group will supply chips worth 750 billion USD to US tech companies like Nvidia. These are framework agreements, not officially signed procurement contracts. But even so, the sheer scale of these figures is breathtaking—equivalent to over half of South Korea's annual GDP. Why is South Korea making such a big move? The answer is simple: positioning for the AI era. HBM high-bandwidth memory is a necessity for AI chips. The global manufacturers capable of mass-producing high-end HBM are essentially SK Hynix, Samsung, and Micron, with South Korea holding two of the three slots. Jensen Huang said HBM is the high-octane fuel for the AI era, and South Korea is directly positioning itself as the 'Saudi Arabia of the AI era.' Saudi Arabia has oil; South Korea has memory chips. However, the market reaction was interesting. After the announcement, the Korean stock market did not surge as expected. This suggests the market is no longer satisfied with 'how big a deal was signed' but is starting to ask 'when will it translate into profits.' After hearing so many stories, everyone wants to see the results. Meanwhile, the portfolio adjustment of the National Pension Service of Korea is also noteworthy. It net sold Korean stocks for six consecutive months in the first half of the year, but in July, it reversed course, turning to net buying, with a focus on increasing holdings of SK Hynix. With the nation locking in deals to expand production and national team funds adding to sector leaders, South Korea's strategic bet on the memory storage track has been fully escalated.

The Big Short's Warning: The Cash Flow Trap Behind the AI Boom. Just as the world celebrates AI and semiconductors, one man has once again positioned himself against the market. Michael Burry, the real-life inspiration for *The Big Short* and the precise predictor of the 2008 subprime crisis, disclosed his latest holdings on Substack on July 25th: increasing his short positions against Micron Technology, Nvidia, and the Philadelphia Semiconductor Index ETF, while adding a new short position against Caterpillar. The specific moves involved shorting more Micron shares at 933.86 USD and adding to his Nvidia short position at 210.28 USD. Burry himself stated that his semiconductor-related shorts now constitute a 'large position' in his portfolio. His core logic can be summed up in one sentence: amidst the current AI infrastructure investment frenzy, a significant portion of the demand does not originate from genuine end customers but circulates within a self-referential loop of opaque financing structures. This judgment is being progressively validated by data. A Reuters calculation based on LSEG consensus estimates suggests that the combined capital expenditure of the five major cloud companies—Microsoft, Google, Amazon, Meta, and Oracle—will surpass their free cash flow for the first time by 2027. The specific numbers are: over the next two years, the five companies' operating cash flow is expected to increase by 340 billion USD, but their capital expenditure needs to increase by 534 billion USD. For every 1 USD earned, they need to spend 1.57 USD. Oracle's situation is the most extreme. In the fiscal year ending May 2026, Oracle's capital expenditure is projected at 55.7 billion USD, against an operating cash flow of only 32 billion USD. It earns 1 USD and spends 1.74 USD on data centers. The company also guided for capital expenditure of 90 to 95 billion USD in fiscal 2027, roughly three times its operating cash flow. This is the greatest prisoner's dilemma of the AI era. Everyone knows AI represents the future, so no one dares to stop. You build a data center, I must build a bigger one. You purchase GPUs, I must place larger orders. You double down on next-generation computing power, I can't afford to fall behind at the starting line. No one dares to be the first to hit the brakes, because stopping could mean being completely left behind in a few years. But when everyone floors the accelerator, the result is that capital expenditure spirals out of control, and the safety margin of cash flow gets thinner and thinner. The day the market discovers that all this computing power can't find enough paying customers, the entire logic will collapse in reverse. This is precisely why Burry dares to go short against the entire market. He isn't betting that AI has no future—he is betting that the pace of AI investment has far outstripped the speed at which real demand can materialize.

In Conclusion: The listing of ChangXin Memory is a landmark moment. It marks the point where China's semiconductor industry finally has a player on the main track of memory storage capable of standing on the global table. From zero to a 7.67% global share, from being choked by foreign technology to achieving self-sufficiency. The journey from its founding in 2016 to its listing in 2026 took a full ten years. It was loss-making for the first nine years, accumulating a total loss of 36.65 billion yuan. It only turned profitable for the first time in 2025. Then, in the first quarter of 2026 alone, it raked in a staggering 24.76 billion yuan in profit. Losing 36.6 billion yuan over a decade, then earning 24.7 billion yuan in a single quarter. This dramatic reversal is more thrilling than any script. It also signifies a profound shift in the market value logic of the A-share market. The once unshakeable 'Big Bank' (ICBC) held the top spot for many years, only to be overtaken by a technology company. The era where resources, finance, and consumption took turns at the helm is giving way to an era of hardcore technological strength. And the story of Country Garden's regret, Alibaba's harvest, Hefei's state capital triumph, South Korea's high-stakes bet, and the Big Short's warning—together, they form the most vivid footnote of this era. Wealth never disappears; it only transfers between different tracks. When an era shifts, the positions of wealth are reshuffled. Some ride the wind to soar, while others are crushed by the wheels of change. There are no permanent winners, nor permanent losers. The only certainty is that those who bet on the next era will ultimately be rewarded by it. And the tales of those who fell before dawn, aside from a sigh, remind every investor: surviving long enough to see your direction pay off is more important than being right about the direction itself.