By | Gold Corner Finance (ID: F-Jinjiao), Author | Tian Yu
Just as ChangXin ignited the imagination of the A-share market, Yangtze Memory arrives.
On August 19th, information from the China Securities Regulatory Commission website showed that the IPO tutoring status of Yangtze Memory Holding Co., Ltd. changed to "tutoring acceptance." The tutoring sponsors are CITIC Securities and China Securities.
Two days later, Yangtze Memory disclosed its prospectus. The most eye-catching set of figures immediately appeared: In Q1 2026, the company's net profit attributable to shareholders reached 33.379 billion yuan, averaging nearly 4 billion yuan earned per day. The money earned in three months is already 2.35 times the full-year amount for 2025, and also exceeds ChangXin Technology's 24.76 billion yuan net profit attributable to shareholders for the same period.
This set of numbers is hard for IPO investors not to get excited about.
Just one month earlier, ChangXin Technology had just listed on the STAR Market. On its first trading day, ChangXin Technology's turnover exceeded 100 billion yuan in less than an hour after opening, with its stock price surging over fivefold at one point. Its closing market capitalization exceeded 3.2 trillion yuan, making it the A-share market's "King of Market Value"; on the fifth trading day after listing, its intraday market capitalization once surged past 4 trillion yuan.
Both are in the memory chip business. ChangXin has already created a wealth myth of 4 trillion yuan; Yangtze Memory's quarterly profit is even higher, and now it also stands at the doorstep of listing.
It seems Yangtze Memory possesses almost all the conditions to become the "next ChangXin."
Can the 4 trillion yuan valuation myth be replicated once more?
The Major Shareholder Collapsed, and the US Caught Up
If you only look at the journey traveled, Yangtze Memory's path has indeed been more bumpy than ChangXin's.
In July 2016, Yangtze Memory Technology Co., Ltd. was formally established. Later, Unisplendour Group held over 51% of the shares, becoming the controlling shareholder, with Zhao Weiguo serving as Chairman and Yang Shining as General Manager.
During those years, Unisplendour Group, helmed by Zhao Weiguo, was at its peak, engaging in numerous mergers and acquisitions, intensive financing, and was once one of the most aggressive capital players in China's tech industry. But as the scale expanded, debt also ballooned.
In July 2021, the Beijing First Intermediate People's Court ruled to accept the bankruptcy reorganization case of Unisplendour Group.
Once Unisplendour collapsed, Yangtze Memory was immediately thrust into a very awkward position.
At that time, the company was still in a critical stage of continuous R&D, capacity expansion, and catching up with international giants, requiring huge amounts of capital annually. After the major shareholder encountered problems, who would take over Unisplendour's equity stake concerned not just financing, but also customer qualification, supply chain cooperation, and even the stability of the core team.
Hubei State Capital ultimately caught it.
By then, Yangtze Memory was already China's only IDM manufacturer capable of mass-producing 3D NAND flash memory. Unisplendour was gone, but Yangtze Memory's factories were still running, R&D hadn't stopped, and the team hadn't dispersed.
In 2022, Hubei Science & Technology Investment Group invested 5.1 billion yuan to acquire assets related to the Yangtze Memory project. By February 2023, Yangtze Memory's registered capital increased from 56.275 billion yuan to 105.27 billion yuan, an increase of nearly 87%. After the capital increase, Hubei Changsheng Development Co., Ltd. became the largest shareholder.
Yangtze Memory finally extricated itself from Unisplendour's debt quagmire, only for US restrictions to follow closely behind.
In December 2022, Yangtze Memory was added to the US Department of Commerce's Entity List, with strict restrictions on the procurement of equipment, materials, and some key components. For a Chinese company still chasing Samsung, SK Hynix, and Micron, not being able to purchase key equipment affected not only capacity expansion but also slowed down the next-generation process upgrades.
In January 2024, Yangtze Memory was also added to the US Department of Defense's Chinese Military Companies (CMC) list.
By 2026, US pressure began spreading from equipment and technology all the way to customers.
Previously, informed sources revealed that Apple was testing memory chips from ChangXin Technology and Yangtze Memory and considering using them in multiple product lines like iPhones and MacBooks.
Although the news remained at the testing and consideration stage, it was enough to alert US politicians. Apple is the most important certification benchmark in the global consumer electronics supply chain. Once entering its supply chain, Yangtze Memory would gain not just orders, but also the endorsement of product quality and stability from a global leading customer.
According to a Bloomberg report on July 30, a bipartisan group of US senators publicly pressured Apple, demanding it abandon purchasing chips from ChangXin Technology and Yangtze Memory, and requiring Apple to make a clear commitment by August 21st not to procure products from the two companies.
On August 14th, The Wall Street Journal further reported that the Trump administration also began publicly pressuring. US Commerce Secretary Howard Lutnick stated that the Trump administration does not want Apple to purchase Chinese memory chips.
Before, it was restricting equipment, materials, and processes; now, even customers are being targeted.
This will certainly increase the difficulty for Yangtze Memory to enter the supply chains of major global customers. But if a company had no competitiveness, it typically wouldn't be chased and restricted all the way from equipment to customers.
In 2016, Yangtze Memory was still striving for the mass production of 3D NAND. Ten years later, the US government already needed to publicly urge Apple not to purchase its chips.
From Unisplendour's bankruptcy, the change of controlling shareholder, to the continuous escalation of US restrictions, Yangtze Memory has been sailing against the wind almost the entire way, yet it was neither dragged down by Unisplendour nor did it stop in its tracks amidst the blockade.
This experience is certainly valuable, but four trillion yuan cannot rely solely on having endured hardship before finding success.
Yangtze Memory has already proven it's hard to defeat. Next, it needs to prove that the current 33.379 billion yuan quarterly profit isn't just catching a favorable wind.
Scale Ranked Third Globally, Revenue Fifth
Yangtze Memory arrives at the doorstep of listing at an almost impossibly better time.
From 2023 to 2025, the company's operating revenues were 18.744 billion yuan, 45.203 billion yuan, and 63.185 billion yuan respectively; net profit attributable to shareholders turned from a loss of 19.181 billion yuan to profits of 6.771 billion yuan and 14.211 billion yuan.
By Q1 2026, Yangtze Memory achieved operating revenue of 47.042 billion yuan and net profit attributable to shareholders of 33.379 billion yuan. In just three months, it earned 2.35 times the profit of the entire year 2025.
But this profit surge cannot be discussed without mentioning price increases.
That quarter, the average selling price of Yangtze Memory's NAND products increased by approximately 173% compared to the 2025 average price, driving the comprehensive gross profit margin to soar from 35.3% in 2025 to 76.8%. Technological progress, yield improvement, and capacity release certainly all played a role, but facing a 173% price increase, even the most sophisticated business analysis can hardly erase the cycle from the profit statement.
Beyond price increases, Yangtze Memory's market share is also rising rapidly.
Data released by Counterpoint Research on August 12 shows that in Q2 2026, Yangtze Memory's global NAND bit shipment share reached 14%, rising to third place globally, behind only Samsung Electronics' 25% and SK Hynix's 22%, and slightly surpassing Kioxia and Micron.
That quarter, Yangtze Memory's shipment volume increased 22% year-on-year and 5% quarter-on-quarter. With the gradual ramp-up of its third wafer fab, Fab 3, Counterpoint expects its third-place global position may even be further solidified by 2027.
This already looks like the profile a global memory giant should have.
But using a different statistical metric, the gap reappears.
According to the same Counterpoint data, while Yangtze Memory's bit shipment volume ranks third globally, calculated by revenue, it still only ranks fifth.
The so-called bit shipment volume can be simply understood as how much storage capacity is sold. But the same 1TB, installed in a regular consumer-grade solid-state drive versus an enterprise-grade solid-state drive for data centers, commands completely different prices.
The NAND that Yangtze Memory currently sells is still primarily consumer-grade products used in phones, computers, and regular SSDs. Plenty of goods are sold, but the unit price hasn't reached the level of enterprise-grade products.
This is also the most realistic situation Yangtze Memory faces today: The scale has caught up to the top three globally, but the product structure hasn't fully kept pace.
During an industry upturn, this problem isn't prominent. As long as the market faces shortages and prices rise, available capacity can turn into profit. But once supply and demand reverse, consumer-grade products are often the first to see price cuts, and profits are impacted more quickly.
This is the nature of the memory chip business. When prices rise, wafer fabs are like printing presses; when prices fall, depreciation, inventory, and massive capacity collectively come to collect their dues.
Therefore, Yangtze Memory is accelerating its transition towards enterprise-grade SSDs. The market expects that by the end of 2026, enterprise-grade SSDs may account for about half of the company's shipments.
This step cannot be delayed.
Counterpoint data shows that in Q2 2026, enterprise-grade SSDs already accounted for 48% of global NAND bit shipments, up from only 26% a year ago. As AI data centers expand their procurement, the profit center of the NAND market is rapidly shifting towards enterprise-grade products.
TrendForce data also shows that in Q1 2026, the revenue of the top five enterprise SSD brands increased 86.1% quarter-on-quarter, with contract prices rising about 80%.
The problem is, the window for Yangtze Memory to complete this transformation may not be very long.
TrendForce predicts that the global NAND market will still face supply shortages in 2026, but by 2027, the supply-demand relationship may begin to loosen. Counterpoint, meanwhile, expects 2028 could become the starting point for NAND prices re-entering a downward phase. Jefferies even warns that the memory price peak may come earlier than the market expects.
For Yangtze Memory, the current golden window is also a countdown.
This IPO is precisely to race against time. During the reporting period, Yangtze Memory's cumulative R&D expenditure was about 15.953 billion yuan, and cumulative cash expenditure for purchasing and constructing long-term assets was about 96.39 billion yuan. Expanding capacity, upgrading processes, developing enterprise-grade products, and then completing major customer qualifications—each step requires continued heavy investment.
According to the plan, Yangtze Memory will use 20.8 billion yuan of the raised funds for mass production line technology upgrades, and another 12.2 billion yuan for R&D projects.
With ChangXin Technology's wealth effect preceding it, raising this 33 billion yuan will likely not be difficult for Yangtze Memory.
The real difficulty is turning this money into enterprise-grade products, customer qualifications, and higher unit prices before NAND prices reverse direction.
Profits Exceed ChangXin, Valuation May Not
Yangtze Memory's Q1 profit exceeded ChangXin's, making it easy for the capital market to follow these numbers with calculations:
If ChangXin can be worth 4 trillion yuan, why can't Yangtze Memory?
Similar expectations have already emerged.
A public fund manager told the media that Yangtze Memory's valuation might be similar to ChangXin Technology's, reaching at least two to three trillion yuan. A private equity practitioner put it more bluntly: These two are both "pure-blooded hard tech." As long as you can get an allotment, you'd wake up laughing in the middle of the night.
But a 4 trillion yuan valuation clearly cannot be based solely on who earned more in a single quarter.
Although both are called memory chips, ChangXin Technology makes DRAM, while Yangtze Memory makes NAND. Their positions on the AI industry chain are not the same.
Simply put, DRAM is more like a server's workbench, responsible for temporarily calling and processing data; NAND is more like a warehouse, responsible for storing data long-term. The larger the model and the more parameters, the server first needs a larger "workbench" to process more data simultaneously.
Therefore, this wave of AI directly drives demand for DRAM more significantly.
In 2025, the global DRAM market size was approximately $150.5 billion, while the NAND market size was about $55.7 billion—the former nearly three times the latter. TrendForce expects that in 2026, general-purpose servers and AI servers combined will consume about 66% of global DRAM.
Some analysis predicts that an AI server may use 8 times the DRAM of a traditional server, while NAND demand is about 3 times that of a traditional server.
NAND is certainly benefiting as well. AI data centers need to store massive amounts of training data, model parameters, and inference results, driving rapid growth in enterprise-grade SSD demand. However, compared to DRAM, which directly impacts server operational efficiency, NAND is not yet the most constrained link in the current AI computing power system.
In other words, when telling the AI story, ChangXin Technology is closer to where shortages are most acute and price increases easiest right now.
Another difference lies in the competitive landscape.
Taking Q4 2025 data as an example, Samsung, SK Hynix, and Micron collectively held about 90.5% of the global DRAM market share. DRAM is essentially a game for the three giants; it's extremely difficult for other players to squeeze in.
ChangXin Technology is China's only DRAM manufacturer that has already formed globally competitive scale. When the capital market priced it, they were buying not just the profits of a chip company, but also the imagination space of being the "Fourth Pole of Chinese DRAM."
The NAND market is much more crowded.
Besides Samsung and SK Hynix, there are major players like Kioxia, Micron, SanDisk, and Yangtze Memory. During the same period, the combined share of the top three global NAND manufacturers was about 64%, far lower than the 90.5% for the top three in DRAM.
Yangtze Memory also possesses the value of import substitution and industrial security, but the scarcity of being the "Chinese NAND Leader" is still different from that of the "Fourth Pole of Chinese DRAM."
What requires more caution is that Yangtze Memory is listing right around the peak of NAND prices and company profits.
In Q1 2026, the company's NAND product average selling price increased about 173% compared to the 2025 average, driving net profit attributable to shareholders to 33.379 billion yuan. If one simply multiplies this single quarter's profit by four and then applies ChangXin's valuation logic, it's easy to mistake a cyclical peak for a long-term norm.
The capital market can certainly pay for import substitution, technological breakthroughs, and future growth. But if the excess profits of a high-growth phase are made permanent, the actual bet is that NAND prices will never fall again.
And the memory industry over the past decades has repeatedly proven that as long as high profits persist long enough, manufacturers will expand production, and supply will eventually catch up with demand. AI might extend this upturn longer, but it hasn't eliminated the cycle.
Yangtze Memory will likely not lack enthusiasm after its listing. It survived Unisplendour's bankruptcy and US containment, reached global third place, and delivered a sufficiently astonishing profit statement.
The capital market can certainly pay for import substitution, technological breakthroughs, and future growth. The question is, as Yangtze Memory lists during the NAND price peak, whether the 33.379 billion yuan should be seen as a new profit level or a special performance during a high-growth phase requires more time for the market to judge.
ChangXin Technology's 4 trillion yuan provides Yangtze Memory with a conspicuous reference point, but it's difficult to become a valuation template that can be directly applied.
When Yangtze Memory truly lists, the A-share market will, for the first time, simultaneously have DRAM and NAND manufacturers with globally competitive scale. By then, the market needs to understand two different businesses separately, rather than looking for the "next ChangXin."
Yangtze Memory's own valuation coordinates will gradually be determined by its progress in enterprise-grade SSDs, global customer structure, and profitability after NAND prices retreat.
How much it rises on its first trading day will be answered in a few hours.
What Yangtze Memory is truly worth needs to be seen after it weathers a full NAND cycle.
References:
- Caixin: "Yangtze Memory Completes IPO Tutoring Acceptance, Listing Valuation Expectations Rise"
- Caixin: "Memory Duo Knocks on the Door of A-Shares"
- National Business Daily: "Yangtze Memory, IPO Progresses Further! NAND Flash Shipment Share Surpasses Micron to Rank Third Globally, Revenue Fifth; Institutions Warn: 2028 May See Price Downturn Inflection Point"
- YQ Industrial Observation: "Why is ChangXin Technology Worth 3 Yangtze Memories?"
- Zero Degree Observation: "The Past of Yangtze Memory"
