Written by: Tide Research

On July 9, Changxin Technology disclosed its prospectus, officially launching the Sci-Tech Innovation Board issuance: July 13 for inquiry, July 16 for subscription, with the stock code 688825. This issuance will offer 6.688 billion new shares, accounting for 10% of the total share capital post-issuance, with no old shares transferred, and if the overallotment option is fully exercised, it could expand to 7.691 billion shares; half of the initial issuance is locked for strategic placement, with executives and core staff subscribing for 10%, while CICC Wealth and CITIC Construction Investment each investing 2%. The company intends to raise 29.5 billion yuan, making it the second-largest IPO in the history of the Sci-Tech Innovation Board, following SMIC.

There is a set of data in the prospectus that reads like two companies.

From 2023 to 2024, the net profit attributable to the parent company is expected to be negative 16.34 billion and negative 7.145 billion yuan, with cumulative unallocated losses of 36.65 billion yuan as of the end of 2025.

In the first quarter of 2026, revenue is expected to be 50.8 billion yuan, a year-on-year increase of 719%, with a net profit of 33 billion yuan, equating to nearly 400 million per day, and a net operating cash flow of 42.57 billion yuan for the quarter.

From abyss to fountain, only one cycle separates them.

The Scale of the Cycle and a Pair of Scissors

DRAM is what we commonly refer to as memory, the chips of the "memory stick" in phones and computers.

The CPU is the brain, the hard drive is the warehouse, and memory is the workbench; when power is lost, it forgets, but without it, nothing runs.

This industry has fought fiercely to date, with Samsung, SK Hynix, and Micron capturing over 90% of the market share. A detail in the prospectus hints at a change in industry leadership: based on 2025 sales, SK Hynix has surpassed Samsung with 34.48% compared to Samsung's 33.96%, marking a leadership change. Changxin ranks fourth globally and first in China with 7.67%, operating three 12-inch wafer fabs in Hefei and Beijing, with capacity utilization rising from 87.06% to 95.73%, nearing full production.

The cruelty of the cycle is summarized by a scale cited in the prospectus from Omdia: from 2015 to 2025, DRAM prices peaked at $7.89/GB, while the low point in the first half of 2023 was $1.78/GB, with a nearly eighty percent difference.

On Changxin's financial statements, the unit price of DDR series products plummeted by 46.61% in 2023, leading to a write-off of 11.5 billion yuan for inventory depreciation over that year, exceeding the annual revenue of 9.1 billion yuan; the depreciation in inventory was greater than the income from sales, illustrating what a bottom cycle looks like.

The top, however, resembles a pair of open scissors.

On the selling price side, by 2025, the unit price of DDR series is expected to rise by 61%, while LPDDR will increase by 24.46%; on the cost side, the unit costs of the two main product lines have decreased by 26.26% and 22.85%, and even the procurement side is helping, with the silicon wafer procurement price index dropping from the base index of 100 to 69.99, and chemicals falling to 73.72. Selling prices rise, buying prices fall, costs dilute, with three forces acting simultaneously, the gross margin has improved from negative 112.71% (pre-reversal) to 37.81%.

By consolidated standards, Changxin's comprehensive gross margin is expected to be 40.99% in 2025, surpassing Samsung's 39.38% and Micron's 39.79%, second only to SK Hynix, which has fully reaped the benefits of HBM. A company that was deeply in the red just two years ago now has a gross margin that matches Samsung’s, which is the most eye-catching data in the entire prospectus.

DDR4 Exit, HBM Absence

Within Changxin's product structure, there are two important signals.

The first signal is written in an inconspicuous paragraph in the fifth section: As of the end of 2024, Changxin has ceased production of its own DDR4 products.

In the past six months, global DDR4 spot prices have risen in rounds, and many analyses attribute this to "Changxin's DDR4 expansion impacting the market."

The facts presented in the prospectus, however, are quite the opposite; it has already exited DDR4 and is fully betting on DDR5 and LPDDR5/5X. The effect is immediate: revenue from DDR series is expected to jump from 3.17 billion yuan in 2024 to 19.53 billion yuan in 2025, a 6.2-fold increase, with its share rising from 13.26% to 31.87%, driven by the rapid release of high-priced and high-margin DDR5. In the overall revenue picture, LPDDR targeting smartphones still accounts for 66.43%, with the top five customers making up 68% of revenue; the end-users include Alibaba Cloud, ByteDance, Tencent, Lenovo, Xiaomi, OPPO, and Vivo.

The second signal is the absence of one term: HBM does not appear anywhere in the text.

Market rumors have suggested that Changxin has been pushing HBM R&D for a long time, yet the application materials make no mention; the 9 billion yuan foresight technology fundraising project is merely described as "research on advanced DRAM-compatible foresight technology," with a construction period from 2026 to 2028. The high-bandwidth memory story for supplying AI servers is not recognized at all in this legal document.

It is worth mentioning that the prospectus also cautiously discusses AI, acknowledging that the revenue share from AI-related fields during the reporting period is relatively low. Changxin benefits indirectly: the three overseas giants are shifting advanced capacity toward HBM, diminishing the supply of general DRAM, which passively lifts prices, and its products just happen to stand at the gap.

An Easily Overlooked Shareholding Chart

A detail regarding equity: among Changxin's three wafer fabs, two are not genuinely "owned" by the group.

For the Changxin XinQiao factory, the group holds only 30.68% directly; for the Beijing factory, Changxin Jidian has a direct holding of 31.72%. Both rely on a concerted action agreement to gather enough voting rights of 73% and 75% for consolidation, with the second phase of the Big Fund directly listed among the shareholders of both subsidiaries, holding 26.99% and 24.67%, respectively.

This structure explains the most glaring gap on the financial statements: in 2025, net profit is expected to be 7.14 billion, but only 1.875 billion belongs to the parent company; in the first quarter of 2026, net profit is 33 billion, with 24.76 billion attributable to the parent company. The majority of the discrepancies belong to the minority shareholders of the two fabs. Investors buying in during the IPO acquire parent company interests at the group level; when looking at the income statement, focus on the parent company line and do not be dazzled by the large figures in net profit.

The Anchoring of the Issue Price is Hidden in the Arithmetic of Strategic Placement

Institutional predictions for Changxin's market capitalization post-listing range from 1 trillion to 4 trillion yuan, but the pricing anchor provided by the prospectus itself is much more sobering, easily seen through simple division.

The first division: the intended fundraising of 29.5 billion yuan divided by 6.688 billion new shares corresponds to an issue price of about 4.4 yuan/share.

The second cross-verification: excluding the employee stock incentive and sponsor co-investment, other strategic investors initially allocated about 2.408 billion shares, with a subscription cap of 10.844 billion yuan, also corresponding to about 4.5 yuan/share.

Calculating based on 4.4 yuan and total share capital of 66.88 billion shares, the issuance market valuation falls within the 300 billion yuan range. The issuance sets at 300 billion while the secondary market starts at 1 trillion, with the gap of more than double in between being the entire space for the bullish-bearish games post-July 16, as well as the source of new share profit expectations under the regulatory practice of "low-price issuance, benefiting the secondary market."

The timeline of the primary market supports this anchor: in June 2025, Alibaba Cloud Computing invested 6.1 billion yuan in Changxin at a price of 2.6302 yuan/share, just thirteen months before the listing. On the founder's side, Chairman Zhu Yiming received 1.536 billion incentive shares at a price of 0.108 yuan per registered capital, committing to distribute half of them to employees within ten years post-listing, locking his personally held shares for 120 months from the listing. A ten-year lock-up is rare in A-shares. The largest shareholder, Qinghui Jidian, has also promised that if the net profit attributable to the parent company after deducting non-recurring items declines by over 50% compared to the previous year, the lock-up period will automatically extend each year.

Risks are Clearly Written in the Financial Statements

The production line supporting a gross margin of 40% is also the heaviest burden on the profit statement.

As of the end of 2025, the book value of fixed assets is expected to be 183 billion yuan, accounting for 54% of total assets; only in 2025, depreciation is expected to be 24.68 billion yuan, 2.3 times that of 2023, and will continue to increase after the fundraising projects are completed. When prices are rising, depreciation can be diluted by profit; when prices fall, it becomes a major source of significant losses like in 2023.

Another number worth monitoring is the production-sales rate: while capacity utilization continues to increase, the production-sales rate has dropped from 99.45% to 90.67%, with nearly 10% of production unsold by 2025, turning into inventory. Stockpiling during a price rise cycle is a smart business, provided that prices continue to rise; the risk section of the prospectus does not mince words, clearly stating that there is a risk that the significant growth of performance in the first half of 2026 may not be sustainable.

Industry expectations for this round of prosperity cycle are generally extended until mid-2027.

Investors who pressed the subscription button on July 16 bought a ticket earned from a decade-long race of China’s storage industry; everyone is curious about how long this storage super cycle can last, but for now, this trillion giant ship has already set sail.