On July 27, 2026, Changxin Technology officially went public on the Science and Technology Innovation Board. According to third-party market capitalization tracking website 8Marketcap, its total market capitalization during trading reached approximately 3.35 trillion yuan, ranking 31st globally among listed companies by market cap. This figure represents a multiple increase compared to its initial valuation of 579.2 billion yuan at the time of issuance. However, the more striking contrast is that, as of the end of 2025, the company had accumulated unrecouped losses amounting to 36.65 billion yuan.
How did a domestic DRAM company, which has experienced a decade of prolonged losses, garner such fervent support in the capital markets? Behind this is not only a core team that returned from overseas and settled in Hefei after ten years of running but also a gamble and harvest from Hefei state-owned assets, brokerages, and the five major banks' AIC and other patient capital.
The "506 Project" and the Logic of the Core Team's Return to Entrepreneurship
The starting point of Changxin Technology can be traced back to May 6, 2016. On that day, Zhu Yiming discussed memory projects with the Hefei municipal government, leading to the birth of the domestic DRAM breakthrough plan codenamed "506." A month later, Changxin Technology was registered and established in Hefei.
Zhu Yiming is not a newcomer to the semiconductor industry. He holds a bachelor's and master's degree in Physics from Tsinghua University and later pursued further studies at the State University of New York at Stony Brook. In 2005, he founded Zhaoyi Innovation in an office in Zhongguancun, Beijing. At that time, the memory chip market was almost entirely monopolized by foreign capital, but Zhu Yiming chose to start with the relatively niche NOR Flash market. Although NOR Flash does not have the market scale of DRAM, its technical barriers remain high and there is a strong demand in embedded devices. With precise judgment on the technological route and continuous research and development, Zhaoyi Innovation gradually tore through the defenses of foreign capital, growing into an important player in the global NOR Flash field. In 2016, Zhaoyi Innovation successfully went public on the A-share market.
The success of Zhaoyi Innovation allowed Zhu Yiming to accumulate deep industry credit capital. He proved that he has the ability to win battles in niche memory sectors and break foreign monopolies. However, DRAM is completely different from NOR Flash. DRAM is known as the crown jewel of the semiconductor industry and has been dominated by three giants: Samsung, SK Hynix, and Micron for nearly forty years. It is a field characterized by both capital intensity and technological intensity, with a high barrier to entry requiring investments on the order of billions, making it nearly impossible for purely market-driven startups to survive. The DRAM market is highly cyclical, and the giants have eliminated countless challengers through brutal price wars during cyclic lows. Around 2016, the global DRAM market was at the start of an upward cycle, driven by the proliferation of smartphones boosting the demand for mobile memory, while the vast domestic terminal manufacturing industry was entirely reliant on imported chips. This mismatch in supply and demand created security risks, becoming the macro backdrop for Changxin's project initiation.
Without the decisive intervention and substantial support from local governments, it would have been nearly impossible to establish heavy assets. The Hefei municipal government extended a helping hand at this time. Prior to this, Hefei had explored an industrial upgrade path through attracting heavy asset projects by introducing BOE, which involved “government attracting heavy asset projects—state-owned capital taking the lead—listing exit—re-investment cycle,” known externally as the "Hefei model." The Hefei municipal government saw the execution ability of Zhu Yiming's team and the strategic value of the DRAM sector, deciding to provide the financial foundation for Changxin's project initiation. For Hefei, bringing in Changxin was not only for financial returns but also to occupy a core node in the semiconductor industry chain and drive the clustering of upstream and downstream supporting enterprises.
In 2018, Zhu Yiming resigned as general manager of Zhaoyi Innovation to officially serve as the CEO of Changxin Technology. He pledged not to take a penny in salary or bonuses before Changxin became profitable. This decisive stance tightly binds technological idealism with local industrial demands. Furthermore, Zhu Yiming also allocated 768 million shares to employees, which, based on the issuance price of 8.66 yuan, amounts to over 20 billion yuan in equity incentives. He himself promised not to transfer shares in the first ten years after the IPO. These details indicate that the birth of Changxin is not merely a market arbitrage act but rather a long race placing his personal reputation and wealth on the line.
Ten Years of Losses and Technical Breakthroughs: The Difficult Ascent from 0 to 1
The barriers in the DRAM industry lie not only in the massive capital investment but also in the extremely complex process adjustments and vast patent blockades. Giants like Samsung have repeatedly used cyclical downturns to wage price wars, eliminating competitors. During the downturns in the storage industry, product prices often fall below cash costs, forcing high-cost production capacity to exit the market. For Changxin Technology to carve a niche in this domain, it must bear vast trial-and-error costs.
The manufacturing processes for DRAM involve extremely tiny capacitor and transistor structures, requiring high precision in photolithography, etching, and thin-film deposition. International giants have accumulated a tremendous amount of process data and patent barriers over decades of development, meaning that later entrants, even if they understand the principles, cannot bypass these patent restrictions. In the early stages of development, Changxin faced comprehensive challenges from process design to yield improvement. To resolve patent issues, Changxin gradually built its intellectual property system through a combination of independent research and development and partial technology introduction. Every failure during the production line adjustment meant that hundreds of millions of funds were wasted. Yield ramp-up is a lengthy process, requiring engineers to fine-tune thousands of process parameters from an initial yield of just over ten percent to a final yield of eighty to ninety percent.
The total investment for Changxin Technology's first phase project was about 18 billion yuan, of which Hefei Industrial Investment contributed 14.4 billion yuan, accounting for 80%. In 2018, Changxin Technology's first production line officially began production. In September 2019, Changxin launched its first 10nm 8Gb DDR4 chip and achieved mass production, marking a breakthrough from 0 to 1 for the DRAM industry in mainland China. Since then, Changxin's research and development system has shown its ability to generate revenue. By November 2023, Changxin launched China's first LPDDR5. By 2025, Changxin completed product coverage for DDR5 and LPDDR5/5X.
By the end of 2025, Changxin Technology had a total of 19,298 employees, with R&D personnel accounting for as much as 32.43%. Currently, Changxin Technology has three 12-inch wafer fabs, two in Hefei and one in Beijing. However, the cost of catching up technologically has been staggering. By the end of 2025, Changxin Technology's accumulated unrecouped losses reached 36.65 billion yuan. Among this, the 2024 revenue was 24.178 billion yuan, while net loss was 7.145 billion yuan.
This kind of loss is the inevitable fate of a follower. In the process of catching up in technology, Changxin has paid a huge financial price. Due to lagging processes, slow yield ramp-up, and high unit bit costs, for a long time, Changxin could only rely on price wars and the sentiment of domestic substitution to secure orders. The enormous losses reflect the high trial-and-error costs in the process of technological catch-up, but they also prove that its research and development system possesses the capability for sustained tackling of challenges.
Capital Marathon and The Outburst of Early Investors' Paper Returns
The capital story of Changxin Technology is the ultimate embodiment of patient capital. During the ten-year period of losses, early investors not only did not withdraw but instead continually increased their investment. From the initial project's 18 billion investment to a pre-Series B valuation of 140 billion by 2024, and then to the last round before the IPO with a valuation of 158.4 billion, Changxin's valuation surge coincides with the competition and succession of various capital.
The Hefei state-owned asset system is the biggest winner in this marathon. In the initial project, Hefei Industrial Investment contributed 14.4 billion yuan. Before the IPO, Hefei state-owned assets held approximately 36.79% of the shares. Based on the market capitalization of 3.35 trillion yuan, the corresponding market capitalization for Hefei's state-owned assets exceeds 1.1 trillion yuan. This investment not only achieved a trillion-yuan financial floating profit but also transformed Hefei into a major hub in the domestic semiconductor industry. The head of Hefei Industrial Investment once stated that the investment was in Zhu Yiming as a person. During Changxin's most difficult loss period, Hefei state-owned assets did not rush to exit but chose to continue injecting capital, using time to gain space for industrial upgrading.
In addition to local governments, brokerages and market-oriented institutions also entered early. China Merchants Securities invested 324 million in the angel round and currently holds about 0.84% of the shares, with a considerable paper profit at the current market value. Cornerstone Capital led a 1.2 billion investment in September 2021 and Huadeng International invested nearly 900 million. Chairperson of Cornerstone Capital, Zhang Wei, believes that the Changxin project is extremely challenging, requiring the integration of massive resources, coordination of multiple parties' opinions, and bearing immense pressure; mere enthusiasm is far from sufficient. Doctor Peng Guie from Huadeng International stated that when the industry's conditions become difficult enough that the financial models cannot quantify them, the founder's judgment becomes the most significant model. These market-oriented institutions entered when Changxin was not yet profitable, betting on the inevitable trend of domestic substitution and the execution capability of Zhu Yiming’s team.
In June 2024, when the industry was in a cold winter, the five major banks' AIC collectively entered against the trend at an increase price of 2.61 yuan per share. Compared to the issuance price of 8.66 yuan, the paper gains for the five major banks' AIC have exceeded 230%. This logic of state intent underpinning the semiconductor foundation is vividly reflected in Changxin. The funding from the five major banks' AIC has a long cycle attribute, which aligns highly with the return cycle of the semiconductor industry.
By the time of the last round of financing before the IPO, Alibaba Cloud invested 6.1 billion yuan for a 3.85% stake, and Tencent also holds 1.50%. Internet giants became shareholders of Changxin to ensure the security of the AI computing power supply chain. The baton of capital has passed from local governments to market-oriented institutions, then to industrial capital, each link yielding substantial paper returns.
2026 Q1 Profit Inflection Point and Industrial Position of AI Computing Power Foundation
The reason Changxin Technology was able to achieve a market capitalization of 3.35 trillion yuan on its first day of trading is primarily due to the financial explosion in the first quarter of 2026.
In Q1 2026, Changxin Technology's revenue reached 50.8 billion yuan, a year-on-year increase of 719%; the net profit attributable to the parent was 24.762 billion yuan, with a consolidated net profit of 33.012 billion yuan. This means that Changxin Technology netted approximately 367 million yuan daily in the first quarter. This explosive performance is the result of the resonance between the DRAM price cycle and the demand for AI computing power. At the beginning of 2026, the contract price of DRAM surged by 58% month-on-month to 63%. At the same time, the explosion of AI servers drove the demand for high-bandwidth memory.
The driving mechanism of AI computing power on memory chips is not simply a simultaneous increase in volume and price. AI large model training requires processing massive parameters, which puts extremely high demands on memory bandwidth. Traditional DDR memory cannot meet the throughput demands of GPUs; HBM (High Bandwidth Memory) stacks multiple DRAM chips together through advanced packaging, becoming a key bottleneck in AI computing power. AI servers not only consume a large amount of HBM but also drive demand for standard DDR5 and LPDDR5 because inference endpoints and edge devices similarly require greater memory capacity. This structural shift in demand has directly pulled up the price center of the entire DRAM market.
In the fourth quarter of 2025, Changxin Technology's share of the global DRAM market reached 7.67%, ranking fourth globally and first in China, successfully breaking the monopoly of the three giants. In the field of HBM, Changxin Technology has already delivered 16nm HBM3 samples and plans to mass produce HBM3E in 2027. Although Changxin's progress in HBM lags behind SK Hynix by about 2 to 3 years, this opens a second growth curve for data centers, allowing it to leap from being purely a domestic substitution target to a crucial piece of the AI computing power foundation.
The storage chip industry is undergoing a reconstruction of its business model from bulk commodities to long-term custom agreements. The explosive demand for high bandwidth and low power consumption storage brought about by AI computing power has resulted in storage manufacturers with HBM technology receiving higher valuation premiums. Although Changxin started late, it successfully embarked on the fast track of AI computing power with its basic share in domestic substitution and the expectations for HBM. In 2025, annual revenue reached 61.799 billion yuan, marking its first profit of 1.875 billion yuan, while the net profit in the first half of 2026 is expected to reach between 50 billion and 57 billion yuan, indicating a clear profit inflection point.
Dissection of Causes for the 3.35 Trillion Market Capitalization and Objective Concerns
However, the 3.35 trillion yuan market value includes a very high premium for domestic substitution and AI computing power sentiment.
From a valuation perspective, Changxin Technology's current price-to-earnings ratio exceeds 300 times (compared to 2025 profitability), far higher than the international peer level of 5 to 8 times. The combined market capitalization of the three storage giants—Samsung Electronics, SK Hynix, and Micron—is about 4.1 trillion dollars. Although Changxin Technology’s market capitalization is approaching some giants, it still has a gap in industrial fundamentals.
The main process of Changxin Technology is approximately 17 to 18nm, lagging behind the international leading level by about 1.5 to 2 years. In terms of unit bit costs for DDR5, Changxin is more than 30% higher than the three major leaders. This means that during price wars in the cyclical downturn of storage, Changxin's cost disadvantage will be amplified, and its profit resilience will face severe tests.
In terms of product structure, Changxin is currently still mainly focused on DDR4 and LPDDR5, while international giants have largely shifted their production capacity to DDR5 and HBM. This product generation gap limits Changxin's penetration in the high-end AI server market. In terms of customer base, the top five customers of Changxin Technology account for over 68% of sales, indicating high customer concentration and existing related transactions with Zhaoyi Innovation. In contrast, international giants enjoy a more diversified global customer base. Although this customer structure ensures Changxin's initial shipping volume, it also restricts its bargaining power in the open market. The accumulated unrecouped losses of 36.65 billion yuan will also need to be gradually digested in future profitable years.
Despite the emergence of a profit inflection point, Changxin is essentially still a company with strong cyclical attributes. The capital market's fervent pursuit is based on expectations of infinite growth in AI computing power. However, the long road of technological catch-up will not be shortened due to a short-term surge in market value. Process generation gaps, cost disadvantages, and high valuation risks are objective realities that Changxin must face after reaching the peak of a 3.35 trillion market capitalization.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
