For the first half of the year, projected revenue is RMB 110–120 billion, with net profit of RMB 66–75 billion—marking a dramatic turnaround from years of massive losses to annualized profits in the hundreds of billions, driven by a genuine and powerful DRAM supercycle.

Author and source: 0x9999in1, ME News



TL;DR

• ChangXin Memory Technologies has received approval from the China Securities Regulatory Commission for its STAR Market IPO. The book-building will take place on July 13, the offering announcement will be published on July 15, and subscription begins on July 16. The online code is 787825. This is the largest A-share IPO of 2026 and the second-largest in STAR Market history, following SMIC’s 53.23 billion yuan offering in 2020.
• The planned fundraising amounts to RMB 29.5 billion, with an estimated offering price of approximately RMB 4.41 per share, corresponding to a market capitalization of about RMB 295 billion upon listing. The free-float ratio under the total share capital after issuance is extremely low—only about 22% of the new shares will be tradable on the first day, corresponding to an estimated free-float market capitalization of approximately RMB 6.5 billion.
• For the first half of the year, projected revenue is RMB 110–120 billion, with net profit of RMB 66–75 billion—marking a dramatic turnaround from years of massive losses to annualized profits in the hundreds of billions, driven by a genuine and powerful DRAM supercycle.
• But the sweetness on paper comes from the industry’s bitterness: Samsung, SK Hynix, and Micron have shifted 70%–80% of their advanced capacity toward HBM and DDR5, causing a contraction in general DRAM supply and a sharp price surge—ChangXin is the one pushed to the forefront of this wave.
• Technically, CXMT’s DDR5 cost per bit remains over 30% higher than that of the three major manufacturers; its HBM technology is still at the small-scale mass production stage for HBM2 and only in sample delivery for HBM3, leaving it one true, full-scale production race away from reaching a higher level beyond “the world’s fourth.”
• Going public is not the end—it’s a moment to lay bare the full extent of the domestic storage industry’s resources, ambitions, and risks for the market to see.

A company that has focused on just one thing for ten years is finally going public.

Let’s clarify things first.

On the early morning of July 9, CXMT announced its issuance schedule: book building on July 13, issuance announcement publication on July 15, and subscription on July 16. The online subscription code is 787825, and the offline subscription code is 688825. The offering aims to raise RMB 29.5 billion, making it the largest IPO on the A-share market this year and the second-largest since the launch of the STAR Market seven years ago—behind only SMIC’s RMB 53.23 billion offering in 2020.

According to the data in the prospectus, ChangXin Technology is initially issuing 6.688 billion shares, accounting for approximately 10% of the total shares after the offering. With a fundraising amount of RMB 29.5 billion, the offering price is approximately RMB 4.41 per share, implying a estimated total market capitalization of nearly RMB 295 billion upon listing.

A stock price under 5 yuan, a market capitalization nearing 300 billion—this is the true scale of China's semiconductor industry.

Now let’s examine the structure. The initial strategic allocation has been raised to the maximum of 50.00%. The offline allocation follows a "3+7" structure, with 70% of the offline allotment shares subject to a six-month lock-up. Calculations show that approximately 78% of the total new shares issued are locked up on the first day, leaving only 22% available for trading on the listing day, corresponding to a free-float market capitalization of approximately RMB 6.5 billion.

A market cap of $300 billion with a floating supply of $6.5 billion.

The two words "supply and demand" perfectly capture the potential for the first day.

A transcript that’s hard to understand

Looking at the performance, it feels even more surreal.

Company forecast: Revenue for the first half of 2026 is expected to range from 110 billion to 120 billion yuan, with net profit attributable to parent company shareholders between 66 billion and 75 billion yuan. What does this mean?

According to public reports, ChangXin Memory Technologies recorded annual revenue of RMB 61.8 billion in 2025, barely turning a profit. Just six months later, revenue nearly doubled, and profits surged from "breaking even" directly to "earning in half a year what a mid-sized internet giant makes in a year." First-quarter revenue increased by 719% year-over-year, with net profit reaching RMB 3.3 billion—compared to a net loss of RMB 2.83 billion in the same period last year.

This curve—anyone who has ever seen a financial report would immediately think: it's too steep.

Too steep means two things.

First, the industry cycle has reached its point. Second, the market will have serious doubts about the "sustainability" of this earnings report.

First, let’s talk about the cycle. This year, the DRAM market has experienced a textbook-level surge. Samsung is renegotiating DRAM contract prices for Q3 2026 with customers, aiming for a further 20% increase compared to Q2. Even in Q2 alone, Samsung’s average DRAM price rose by 50%–60% quarter-over-quarter, and in Q1, it surged approximately 90% quarter-over-quarter.

Why is the price surging so sharply? Because the supply side has collectively defected. Samsung, SK Hynix, and Micron have shifted 70%–80% of their advanced process capacity or new capital expenditures toward HBM and DDR5/LPDDR5X. Together, these three companies control approximately 90% of the global DRAM market—in Q1 2026, Samsung held about 38%, SK Hynix about 29%, and Micron about 22%. If they stop producing general-purpose DRAM, what happens to the rest of the market? Only price increases—and bringing in substitutes.

Changxin is the backup, and the only one with sufficient scale to step in.

Let’s talk about sustainability. According to TrendForce data, in the first quarter of 2026, the wafer revenue and profitability of 64GB DDR5 RDIMM server memory modules surpassed those of HBM products for the first time. This means that the most profitable storage product today is no longer HBM, which is closely tied to AI GPUs, but rather high-capacity DDR5, which is being driven by supply shortages.

ChangXin's unit bit cost for DDR5 is still more than 30% higher than that of the three major manufacturers. However, due to exceptionally strong DDR5 pricing in the first quarter, ChangXin's gross margin was pushed above 70%. According to SemiAnalysis's estimates, by the first quarter of 2026, ChangXin's average selling price for DRAM will be only 5%–10% lower than that of the three major players—not 50%, but just 5% to 10%.

The impressive performance of this semi-annual report stems from the combined benefits of industry-wide structural shortages, domestic substitution, and high-priced general-purpose DRAM. While these advantages are real, no one can guarantee they will be sustained in the long term.

From "technology follower" to "cycle beneficiary"—what did ChangXin get right?

The interesting part is here.

ChangXin was founded in 2016 and only began small-scale shipments of DDR4 in 2019, earning teasing from peers for "doing in five years what others took twenty years to accomplish." Even as of 2023, the company was still operating at a loss, with financial figures remaining unimpressive.

At that time, the market consensus was that China’s breakthrough in DRAM was bound to be a long-term battle—process advancement required time, yield improvement required time, customer certification required time, and cost reduction depended even more on time.

Looking back now, ChangXin did three things right.

First, it chose the right segment for the "second half." While the three major players are fully focused on HBM, CXMT did not directly compete in HBM; instead, it concentrated on mainstream products like DDR4 and DDR5, as well as niche markets. It took on the tasks that others didn’t want to do but couldn’t avoid.

Second, capacity expansion has not stopped. In 2025, monthly production is expected to reach approximately 200,000 wafers, with a plan to increase to 500,000 wafers per month by 2028. SemiAnalysis estimates that by the end of 2026, CXMT’s monthly capacity will reach 350 k wafers per month (kwspm), nearing Micron’s 385 kwspm, and potentially ranking as the third-largest globally by wafer capacity alone. Simply based on this capacity trajectory, CXMT has transformed from a "supporting player" into a "key variable" in global DRAM supply.

Third, it has coincided with an AI supply gap. Over the past three decades, the DRAM industry has experienced multiple cycles, but none have seen suppliers actively abandoning the general market—as is happening now. This is not something CXMT earned through effort; it was handed to it by historical circumstance.

The question is—how long can you hold the gift history has given you?

22% of the circulating supply: a carefully orchestrated illusion of scarcity

Return to the issuance structure.

The initial strategic allocation has been maximized at 50.00%, the upper limit permitted by the STAR Market. The offline allocation is locked for 70% under the "3+7" structure, and the portion held by major pre-IPO shareholders with more than 51% ownership is locked for 36 months. The largest shareholder, Qinghui Integrated Circuit, has added an additional condition: if, starting from the listing year, the non-GAAP net profit attributable to parent company shareholders declines by more than 50% compared to the year prior to listing, the lock-up period will automatically be extended by an additional 12 months.

The lead underwriter, CICC, holds a 15% over-allotment option (commonly known as the "greenshoe"). Within 30 calendar days after listing, if the stock price falls below the offering price, CICC may purchase shares on the secondary market to stabilize the price.

Looking at these points together makes the logic clear.

On the first day, 22% of the circulating supply, equivalent to a market cap of 6.5 billion, corresponds to a company with a nearly 300-billion market cap. Scarcity is engineered by design.

The greenshoe provides a cushion against downward price movement.

Major shareholders have locked in their holdings deep, sending a clear message to the market: I’m not leaving, so you can hold with confidence.

The profit-guaranteed lock-up extension clause ties the "performance realization" to the major shareholder's unlock schedule—lock for three years if performance does not drop by more than half; extend by an additional year if performance drops by more than half.

This combination of measures reveals a shared intention between the issuer and regulators: to raise the full amount in one go while avoiding excessive price volatility on the first day and during the initial trading period.

In a sense, ChangXin’s pricing and structure have given domestic semiconductor companies a dignified opportunity to enter the market. Whether they can go far on their own after that is another story.

Entry barrier for the game: A practical reminder for retail investors

Let’s start by putting things in perspective—this coin isn’t something you can just pump at will.

Online retail investors applying for (787825) must simultaneously meet the following requirements: activate trading rights for the STAR Market and hold an average daily market value of over RMB 10,000 in Shanghai-listed stocks before July 14.

Institutions participating in offline inquiries face stricter requirements: a minimum of RMB 60 million in Shanghai Stock Exchange market value, along with an average daily holding of over RMB 6 million in the STAR Market.

In other words, the capital that makes it to this table has already been filtered. Retail investors can participate in new offerings, but their chances of securing shares—based on recent popular large-cap stocks on the STAR Market—are注定 extremely low, akin to a lottery with minimal odds.

Therefore, the博弈 in the secondary market cannot avoid these key variables:

First, can the initial circulating market cap of 6.5 billion support a $300 billion valuation? Scarcity implies elasticity, and elasticity means potential for sharp movements in either direction.

Second, will the unlocking of 70% of the private placement shares after six months become the first "speed bump"?

Third, the trend of non-GAAP net profit attributable to shareholders in the year of listing—this not only affects the lock-up period of major shareholders but also directly determines how the market prices Changxin as either a "cyclical growth stock" or a "cyclical stock."

Fourth, external policy variables. U.S. export controls on semiconductors to China continue to intensify; ChangXin is already on the U.S. Entity List, and any minor disturbance in the supply chain is transmitted to sentiment in the secondary market.

What did this listing solve, and what did it leave behind?

First, let's address what was resolved.

A fundraising of RMB 29.5 billion, combined with support from major shareholders, ensures ChangXin’s capital expenditures for the coming years are largely secured. According to the prospectus, the funds will primarily be allocated to projects including the technological upgrade of memory wafer manufacturing production lines, DRAM technology enhancement, and advanced DRAM research and development. Of this, RMB 13 billion is designated for memory technology upgrades and RMB 7.5 billion for production line modifications—both are direct, substantial investments in capacity and process capabilities.

Getting the money is just the first step. The real significance lies in Changxin transforming from a project company backed by "national will, local capital, and industry funds" into a truly public company. Disclosure of information, corporate governance, and financial compliance must all withstand market scrutiny. This is an essential lesson for China’s semiconductor industry as it matures.

Again, what remains?

First, the cost gap of DDR5. The unit bit cost being over 30% higher is a reality that domestic DRAM cannot avoid in the short term. The greater the benefits from the current cycle, the more destructive this gap will be when the cycle reverses.

Second, the progress on HBM. Public information indicates that CXMT has begun small-scale mass production of HBM2 and is sampling HBM3 to leading AI manufacturers, with plans for low-volume pilot production by the end of 2026 and full-scale mass production in 2027. The new Shanghai facility includes a dedicated production line capable of manufacturing 50,000 HBM3 wafers per month. The direction is correct, and the timeline is on track; however, compared to SK Hynix and Samsung’s progress on HBM3E and HBM4, CXMT remains in a "follower" position.

Third, customer structure. DRAM is a business with highly concentrated customers. Names like Apple, Amazon, Google, Microsoft, Meta, and domestic companies such as ByteDance, Alibaba, Tencent, and Huawei—how deeply they are integrated into the customer list directly determines whether CXMT can upgrade from a "backup" to a "primary" supplier. Reports suggest that Apple has begun including CXMT as an option during the global DRAM shortage; such signals are significant, but they require sustained time to materialize, not just one or two orders.

Fourth, a valuation anchor. Domestic DRAM now has a public price. After listing, the market will answer a critical question with real money—how much are China’s storage assets truly worth? This price will, in turn, reshape the valuation benchmarks for Yangtze Memory, Hefei JHSC, and all companies along the storage supply chain.

A phrase you can keep tucked away in your heart

At this point, I’m tempted to quote a familiar saying: This is not an ordinary IPO.

But then again, I thought it wasn't necessary.

For ChangXin, going public is just one small step in a much larger mission. What’s truly difficult has never been passing the review, issuing shares, or how much the stock rises on the first day—it’s every quarter ahead, every chip, every customer certification, and every process iteration that follows.

For the market, ChangXin is a mirror. Looking at it, you see where the global memory cycle stands; looking at it, you see how far domestic substitution has progressed; looking at it, you also see how the A-share market values a company that is still catching up technologically but racing ahead financially.

The three major manufacturers raised prices sharply because they know AI demand is real; ChangXin dared to go public at this time because it knows the opportunity window is real.

As for how long the window will remain open, the industry won’t tell you, and ChangXin won’t tell you either.

Only time will tell.

Reference materials

1. Announcement of the Results of the 9th Review Meeting of the Science and Technology Innovation Board Listing Committee of the Shanghai Stock Exchange, May 27, 2026.
2. Approval from the China Securities Regulatory Commission on the Registration of the Initial Public Offering of Changxin Technology Group Co., Ltd., June 12, 2026.
3. Changxin Technology Group Co., Ltd. "Prospectus for the Initial Public Offering and Listing on the STAR Market" and "Announcement on Issuance Arrangements and Initial Inquiry," July 2026.
4. Reuters, "China's CXMT to start book-building on July 15 for $4.3 billion Shanghai IPO," July 9, 2026.
5. South China Morning Post (SCMP), "Inside CXMT's $4.3B IPO: Soaring Profits Meet U.S. Export Threat and High-Stakes HBM Race," June 2026.
6. TrendForce, Q1 2026 DRAM and HBM Market Research Report.
7. SemiAnalysis, "China Memory: CXMT Capacity, Cost, and Pricing Deep Dive," June 2026.
8. 21st Century Business Herald: "In-Depth Report on ChangXin Technology: From Zero to Catching Up in 10 Years, Aiming for Third Globally!" June 25, 2026.