In recent days, Hefei's ChangXin Technology (CXMT) landed on the A-share market, becoming the most valuable stock by market capitalization on its first trading day, even surpassing Tencent momentarily to become China's highest-valued listed company. Wuhan's Yangtze Memory Technologies (YMTC) is also in the IPO guidance phase.
However, in Jinjiang, Fujian, a small city famous for its athletic shoes, there hides another super storage company—Fujian Jinhua Integrated Circuit Co., Ltd. (Fujian Jinhua).
Once upon a time, Fujian Jinhua was also one of the three major memory chip bases alongside Yangtze Memory and ChangXin Technology. In its early stages of establishment, it even left ChangXin Technology behind and was once the domestic memory manufacturer most favored by the industry.
It's just that Fujian Jinhua's script wasn't as smooth. Before its products arrived, it faced the U.S. Department of Commerce's Entity List and criminal charges from the Department of Justice. External upheavals brought Fujian Jinhua's production lines to an immediate halt, and for the next five years, it almost vanished from public view.
On February 27, 2024, a not-guilty verdict from a U.S. federal court in San Francisco brought this long-silent company back into the spotlight. It had been exactly eight years since Jinhua's founding. During these eight years, the global DRAM market also experienced cyclical changes and entered an AI-driven super boom cycle. Needless to say, giants like Samsung and SK Hynix have already made immense profits.
On July 27, ChangXin Technology captured the title of A-share king. By that time, Jinhua had only been free from charges for just over two years.
An unofficial figure from last year showed that Fujian Jinhua's valuation was approximately 80 billion yuan. The accuracy of this valuation is unknown as it hasn't raised capital for a long time. However, although Fujian Jinhua's main business is DRAM, primarily customized DDR4 chips—not the DDR5 and HBM relied upon by AI—the successful listing of ChangXin Technology set a good example. With memory chip prices continuing to rise, Fujian Jinhua can at least ride this wave and benefit somewhat.
Dreams and Controversy
Just as the story of ChangXin Technology is inseparable from Zhu Yiming, Jinhua also has a soul figure—Chen Zhengkun.
He keeps a low profile, rarely gives interviews, and there isn't even a detailed verified biography of Chen Zhengkun available in the market.
Limited information indicates he was born after 1965, showing an interest in semiconductors from a young age. He studied at National Tsing Hua University in Taiwan and the University of California, Berkeley. After graduation, Chen stayed in Silicon Valley working as a chip engineer. Later, he returned to Taiwan and joined an electronics company called Rexchip. In 2013, Rexchip was acquired by Micron, and Chen became the president of Micron's Taiwan subsidiary, which was responsible for manufacturing 25nm DRAM chips for Micron.
During his time at Micron, Chen Zhengkun increasingly saw a clear fact: no matter how high a Chinese engineer rose in management, they would never touch the true core technology.
In 2015, he formally resigned from Micron and joined United Microelectronics Corporation (UMC).
On the other side, on February 26, 2016, Fujian Jinhua Integrated Circuit Co., Ltd. was registered in Jinjiang City, with registered capital of 34.477 billion yuan. It was jointly funded by state-owned enterprises at the provincial, municipal, and county levels, including Fujian Electronics and Information Group, Quanzhou Financial Holding Group Co., Ltd., and Fujian Jinjiang Industrial Development Investment Group Co., Ltd.
This was an enterprise born with a national mission. It was included in China's "13th Five-Year Plan" for major integrated circuit productivity layout, with the sole goal of tearing open a gap in the DRAM field and breaking the decades-long monopoly held by three overseas giants: Samsung, SK Hynix, and Micron.
At that time, there were three memory projects nationwide: Wuhan's Yangtze Memory focused on NAND flash, Hefei's ChangXin independently developed DRAM, while Fujian Jinhua planned to develop niche DRAM through cooperation with Taiwan's UMC.
Therefore, UMC signed a technology cooperation agreement with Fujian Jinhua. Jinhua would be responsible for investing funds to build production lines and purchase equipment, while UMC would send a core technical team to jointly develop 32-nanometer process DRAM technology. Jinhua provided $300 million for purchasing R&D equipment and paid UMC $400 million according to progress. The developed technology achievements would be shared by both parties.
In February 2017, Chen Zhengkun took office as the General Manager of Jinhua. According to multiple media reports, Chen hesitated almost not at all upon receiving the invitation. It's worth noting that what he gave up was not just a high salary at Micron. At that time, Jinhua didn't even have a decent R&D building; the temporary office was set up in an idle factory building next to the industrial park, with a construction site outside the window.
When asked why he joined the UMC and mainland China DRAM technology cooperation and R&D plan, Chen Zhengkun stated that the acquisition of Rexchip by Micron had a huge impact on him back then. Independently developing DRAM technology had always been a dream in his heart.
With Chen Zhengkun at the helm, the speed of cooperation progressed beyond expectations. The factory construction took less than two years.
According to the plan, the first-phase project had a total investment of $5.3 billion, scheduled for official production in the third quarter of 2018. The monthly capacity for 12-inch wafers using the 32nm process was planned to reach 60,000 pieces, with the ultimate goal of launching 20nm products. By 2025, after four phases of construction, the monthly capacity would reach 240,000 pieces.
Everything seemed on track.
Then, the storm came. In September 2017, Micron sued UMC in Taiwan, accusing employees who jumped from Micron to UMC of stealing Micron's DRAM trade secrets, allegedly helping UMC develop 32nm DRAM. In December of the same year, Micron sued Jinhua and UMC in a U.S. federal court in California. Jinhua quickly counterattacked, suing Micron for patent infringement in its sales of products in China, and eventually won.
Unexpectedly, this was just the prelude. In late October 2018, the U.S. Department of Commerce, citing national security reasons, announced the addition of Fujian Jinhua to the Entity List for export controls. This meant not only could U.S. companies not do business with Jinhua, but even global suppliers using U.S. technology were prohibited from supplying it. More seriously, Jinhua's newly built production line, which relied on the global supply chain, was cut off by this ban.
Just one month later, the U.S. Department of Justice formally indicted Fujian Jinhua, UMC, and Chen Zhengkun, among others, escalating the original civil dispute into a criminal case. Charges included conspiracy to commit economic espionage.
In October 2020, UMC chose to compromise, reaching a settlement with the U.S. Department of Justice, admitting to infringing a trade secret, agreeing to pay a $60 million fine, and promising to cooperate with the U.S. government's investigation of Jinhua during a three-year probationary period. Two months later, a U.S. court issued a global arrest warrant for Chen Zhengkun.
However, as neither mainland China nor Taiwan has an extradition treaty with the United States, Chen Zhengkun remained in the country continuing to lead R&D.
The turning point appeared at the end of 2023 when Micron announced a global settlement agreement with Fujian Jinhua, with both parties withdrawing their lawsuits against each other worldwide, ending all litigation. Two months later, on February 27, 2024, a U.S. federal court in San Francisco ruled that prosecutors failed to prove Jinhua misappropriated Micron's proprietary data. All charges were not guilty, clearing Fujian Jinhua's name.
Although an innocent outcome was ultimately achieved, it must be acknowledged that this six-year-long accusation severely slowed Jinhua's progress. Meanwhile, in January 2025, Chen Zhengkun stepped down as General Manager of Jinhua upon the expiration of his employment contract, transitioning to a technical advisor role for the company. By then, he was already in his sixties.
The Hardest Chip to Make
To understand the situation Jinhua faces, one must first understand the DRAM industry.
The essence of DRAM is a super-large array of storage cells. Each cell contains a transistor and a capacitor. The transistor can be made smaller and smaller, but shrinking the capacitor is extremely difficult. When the capacitor becomes small, it struggles to retain charge stably, easily losing it or being interfered with by adjacent components.
To make DRAM more efficient, manufacturers can only attempt increasingly complex structures, making manufacturing extremely expensive. Today, building a state-of-the-art DRAM wafer fab easily requires an investment of $15-20 billion, not to mention the additional billions for purchasing equipment like lithography and etching machines.
More crucially, DRAM is a commodity product, meaning it follows the same set of industry standards. Chips produced by any manufacturer can be directly inserted into any other manufacturer's equipment.
Looking at the history of DRAM development, it's a cycle of booms and busts, a cycle of life and death.
For example, in the 1970s, Intel dominated with a market share reaching 82.9%. In the 1980s, Japan's NEC, Hitachi, and Toshiba staged a comeback with low prices, surpassing 50% market share in 1985, causing a full retreat for the U.S. During the same period, South Korea's Samsung acquired Korea Semiconductor and officially entered the DRAM market in 1983.
Then, after 1990, the industry entered a downturn. During the winter when the entire industry was losing money and DRAM prices plummeted, Samsung instead went on a crazy debt-fueled expansion spree, pushing costs below its competitors' cash flow break-even point through economies of scale. In 1998, Korean companies' market share surpassed Japan's for the first time. In 2001, Hyundai split and renamed Hynix, while Micron gradually grew through continuous acquisitions.
By the time of the 2008 financial crisis, the memory industry faced another major reshuffle. Germany's Qimonda and Japan's Elpida Memory successively declared bankruptcy, marking the exit of Europe and Japan. Afterwards, the global DRAM market eventually converged into an oligopoly pattern dominated by the triumvirate of Samsung, SK Hynix, and Micron, collectively holding over 95% market share.
The surviving giants reached a subtle oligopolistic默契, no longer easily launching all-out price wars. Competition shifted focus to technological positioning. Like the recent hype around HBM (High Bandwidth Memory), HBM involves not only process scaling but also tests advanced stacking and packaging technologies, including 3D stacking and Through-Silicon Via (TSV) technology. Currently, Samsung and SK Hynix have established technological barriers in these cutting-edge fields.
Looking domestically, when Qimonda went bankrupt, local state-owned assets in China attempted to acquire all its assets. However, amid various internal and external interventions and博弈, it took several years and ultimately did not succeed.
Another reality is that around 2015, chips surpassed crude oil imports for the first time to become China's largest imported commodity. At that time, importing memory chips into China meant prices were entirely dictated by Korean and American manufacturers, and purchases were subject to quota restrictions. Domestic memory chip production was almost zero, with no autonomy over the entire industrial chain.
The reason why Fujian Jinhua attracted Micron's attention was, on one hand, the fact that many technical experts and chip engineers joined Jinhua. On the other hand, in 2016, over 40% of Micron's annual revenue of $12.4 billion came from memory procurement in mainland China. DRAM and NAND flash together contributed over 90% of the company's sales. It's no exaggeration to say that the Chinese market became the foundation of Micron's performance.
This high degree of单一 market dependence also laid the commercial motivation for Micron's subsequent launch of cross-border intellectual property litigation and pushing for Entity List restrictions against Jinhua.
In comparison, in its early stages, ChangXin Technology also absorbed several employees from Samsung, with some cases prompting South Korean judicial intervention. Simultaneously, it spent heavily to purchase Qimonda's patent portfolio and signed a patent licensing agreement with the U.S.'s Rambus, obtaining licenses to implement a large number of DRAM technology patents.
Although the price seemed hefty at the time, it successfully circumvented the technical patent issues of Micron, Samsung, and SK Hynix. Later, Zhu Yiming revealed in 2019 that ChangXin Technology had modified some technologies, completely eliminating U.S.-related technology.
Rebuilding from Desperation
After being placed on the Entity List, Jinhua faced an almost impossible task: rebuilding a functional DRAM production line without American equipment and material supplies.
Here, we must mention an experience of Chen Zhengkun.
During the 2008 financial crisis, Chen Zhengkun was serving as General Manager of Taiwan's Rexchip Electronics. At that time, the company had a deep cooperation with Micron and possessed a technologically advanced DRAM production line. However, the yield rate of wafers from this line dropped to just over 60%. Calculated, producing each wafer resulted in a loss. Senior management was already discussing whether to shut down the line directly and sell the equipment as scrap.
Chen Zhengkun, leading a technical team, stayed on that production line for four months, re-optimizing various data parameters, and eventually increased the yield rate to over 80%.
The line that was previously losing $3 million per month turned into a profit of $2 million per month, holding up the company during its most difficult time. After Rexchip was acquired by Micron, this experience became a legendary chapter in his career. Within the company, his name became associated with the nickname "Yield Magician."
There are rumors that initially, about 70% of Jinhua's equipment relied on U.S. supplies. After being placed on the Entity List, Chen Zhengkun led the team on the front lines, boldly overhauling domestic equipment, thereby reconstructing the entire production process logic. This was a long and painful process. However, precisely with this tenacity, Jinhua significantly compressed the proportion of U.S. technology in its equipment.
On Jinhua's official website, there is a page clearly recording "Jinhua Milestones." Here, we see no noisy disputes or controversies, only product progress and revenue breakthroughs.
Behind these numbers lies the slow recovery of an enterprise under extreme pressure.
Of course, there is still a considerable distance between these data and contemporarily established ChangXin Technology and Yangtze Memory. ChangXin Storage achieved mass production of 19nm DRAM in 2019, and by Q1 2025, its revenue had reached 50.8 billion yuan. Yangtze Memory took the lead in mass-producing 232-layer 3D NAND in 2022, aiming for a 9% market share by 2025.
Fortunately, Jinhua was only slowed down, not knocked off the gaming table. Additionally, Jinhua深耕 (deeply cultivates) the niche DRAM market, with products widely used in smart TVs, set-top boxes, printers, routers, industrial control equipment, and other fields. Such products typically have a single product lifecycle of five to ten years. Once entering the supply chain, they exhibit极强的 stability.
Currently, Jinhua's 12-inch wafer fab maintains a stable monthly capacity of approximately 40,000 wafers. According to the plan, capacity will be further expanded to 60,000 wafers per month by 2026. The company holds 1007 related patents, with a technology system covering the entire process from chip design to production manufacturing. However, Jinhua remains on the U.S. Entity List. At the end of 2024, the U.S. Department of Commerce further escalated restrictive measures. It's clear that although legal obstacles have been cleared, technological封锁 (blockade) is far from over.
The Jinjiang Experience
When mentioning Jinjiang, what comes to mind first? Jinjiang Literature City, Panpan snacks, or Anta, Xtep? But definitely not chips.
On the map of China's urban economy, Jinjiang is a symbol of China's private economy. Jinjiang's private economy contributes over 90% of the city's GDP, with one in every seven Jinjiang residents being a business owner.
The numbers are more intuitive: Jinjiang-branded men's clothing accounts for about 25% of national production, athletic shoe production accounts for about 20% globally, snack food accounts for about 20% nationally, and exterior wall ceramics account for about 60% nationally. Well-known brands like Anta, Xtep, 361 Degrees, Hongxing Erke, Panpan, Hengan, Lilang, Joeone all come from this small city.
But no one expected that this city, famous for footwear, apparel, and food, would plunge headfirst into the memory chip industry—one of the most capital-intensive and technologically壁垒最高的 (high-barrier) industries globally.
Some say introducing Jinhua was an aggressive industrial transformation for Jinjiang over the past decade. Among the three contemporarily established enterprises, Wuhan had the foundation of Wuhan Xinxin Semiconductor accumulated over a decade, Hefei invested with the strength of the entire province, while Jinjiang had almost zero semiconductor industry foundation before this. Moreover, Jinhua's first-phase investment of 37 billion yuan was almost equivalent to several times Jinjiang's fiscal revenue that year.
Jinhua was the seed player for Jinjiang's integrated circuit industry, so it started the most aggressively, progressed the fastest, and landed most thoroughly.
As supporting measures, in November 2017, the Fujian Provincial Government approved the establishment of the provincial-level Quanzhou Semiconductor High-Tech Industrial Park, consisting of "one district, three parks." The Jinjiang sub-park focused on developing integrated circuits, the Nan'an sub-park on compound semiconductors, and the Anxi sub-park on optoelectronics primarily based on LED manufacturing.
It was this "mobilizing the entire city's strength" investment that enabled Jinhua to maintain operations and wait for a turnaround even during its darkest hours under sanctions and production line shutdowns. Fujian's state-owned assets hold an absolute controlling stake of over 60% in Jinhua. Throughout the整整五年 (entire five-year) frozen低谷期 (low period), there was no withdrawal of investment, no抽贷 (recalling loans), no disbanding of teams, and no停发薪资 (stopping salary payments). The provincial and municipal governments provided全程兜底 (full guarantee throughout), ensuring factory area maintenance, utilities, and continuous supply of industrial resources.
This stems from Jinjiang's clear定位 (positioning) for itself—guiding hand, pushing hand, and service provider. "Do not intervene unless called upon, be ready upon call, deliver on promises. Do not interfere when enterprises can manage well; give a push when they encounter difficulties 'climbing slopes and crossing ridges' in their operations." A government department head once summarized it this way.
Semiconductors are a typical重资本 (capital-intensive) industry. Without sufficient financial support, everything is空谈 (empty talk).
Centering on the goal of building a provincial-level fund集聚区 (cluster zone), Jinjiang successively established 2 industrial mother funds and 12 government-guided funds, forming 2 fund cluster zones with a total scale exceeding 45 billion yuan. In March 2025, Jinjiang又 (again) established a Talent Technology Innovation Fund with a total scale of 500 million yuan, with an initial phase of 50 million yuan, focusing on investments in new quality productive force fields like integrated circuits, nuclear technology applications, and intelligent equipment.
From Jinjiang's perspective, during the起步阶段 (initial stage) of cultivating emerging industries, market-related funds are often unwilling to介入 (intervene). At this time, the government needs to provide前瞻性引导 (forward-looking guidance).
In recent years, with Jinhua as the龙头 (leading dragon head), Jinjiang has successively对接落地 (connected and landed) 52 integrated circuit industry chain projects, with total investments exceeding 100 billion yuan, forming a全链条产业集群 (full-chain industrial cluster) covering chip design, manufacturing, packaging & testing, equipment & materials, supporting facilities, and终端应用 (terminal applications).
In 2024, Jinjiang's integrated circuit industry is expected to achieve annual规上企业产值 (output value of designated scale enterprises) exceeding 10 billion yuan, a year-on-year growth of 51%. In 2025, the规上产值 (designated scale output value) is预计 (expected) to突破 (break through) 14 billion yuan.
Over more than 20 years, Jinjiang City's regional GDP increased from 27.7 billion yuan in 2002 to 336.35 billion yuan in 2023, equivalent to再造了11个晋江 (recreating 11 Jinjiangs). In the first three quarters of 2024, Jinjiang's GDP reached 249.46 billion yuan, a year-on-year increase of 8.2%, ranking first in growth rate among China's top five counties.
It indicates that this city骨子里藏着 (in its bones hides) an不甘人后的倔强 (unyielding stubbornness unwilling to fall behind).
Now, China's memory industry is no longer the任人宰割的局面 (situation of being at others' mercy) from eight years ago. In Q3 2025, the global DRAM market size reached $40.037 billion, setting a new quarterly historical high. Morgan Stanley predicts that this AI-driven super cycle for memory will last for several years, with the global memory market size预计突破 (expected to break through) $300 billion by 2027.
In this super cycle, Jinhua's monthly capacity of 60,000 wafers and annual revenue of 2 billion yuan may seem微不足道 (insignificant), but it represents the fact that an enterprise that had its neck卡过 (choked) has stood up again amidst封锁 (blockade).
Missing one ChangXin does not mean missing the entire era.
This article is from the WeChat public account "Touzhongwang," author: Zhang Xue
