As ChangXin Memory Technologies drew fresh attention after its A-share listing, another Chinese memory maker has quietly re-emerged from the background: Fujian Jinhua Integrated Circuit Co. Ltd., based in Jinjiang, Fujian.
Jinhua was once grouped with Yangtze Memory Technologies and ChangXin as one of China’s three major storage chip bases. In its early days, it was even seen by some in the industry as the most promising domestic memory player. But before its products could scale, the company was hit by a U.S. Commerce Department entity-list designation and later by U.S. criminal charges. Its production line stalled, and the company largely vanished from public view for years.
The turn came on Feb. 27, 2024. A federal court in San Francisco found that prosecutors had failed to prove Fujian Jinhua misappropriated Micron proprietary data, and all charges were rejected. The ruling put the company, founded eight years earlier, back under the spotlight.
On July 27, ChangXin took the top spot in the A-share market by market value. Jinhua, by contrast, had only been free of the case for a little more than two years. An unofficial figure cited in the article put Fujian Jinhua’s valuation at roughly 80 billion yuan last year, though the number is hard to verify because the company has not raised capital in public markets for a long time. Its main business remains DRAM, mostly customized DDR4 products rather than the DDR5 and HBM lines tied more directly to AI demand. Even so, with ChangXin’s listing setting a benchmark and memory-chip prices continuing to rise, Jinhua may still benefit from the broader upcycle.
Chen Zhengkun and the company’s early push
Any account of Fujian Jinhua runs through one person: Chen Zhengkun.
Chen has kept a low public profile and rarely gives interviews. Detailed, broadly verified biographical material is scarce. The article says he was born in the 1960s, showed an early interest in semiconductors, studied physics at National Tsing Hua University in Taiwan and later attended the University of California, Berkeley. After graduation, he stayed in Silicon Valley as a chip engineer, then returned to Taiwan to join Rexchip. When Micron acquired Rexchip in 2013, Chen became president of Micron’s Taiwan subsidiary, which manufactured 25nm DRAM chips for Micron.
In 2015, Chen left Micron and joined United Microelectronics Corp., or UMC.
Fujian Jinhua Integrated Circuit Co. Ltd. was registered in Jinjiang on Feb. 26, 2016, with registered capital of 34.477 billion yuan. Its investors included Fujian Electronic & Information Group, Quanzhou Financial Holding Group Co. Ltd., and Fujian Jinjiang Industrial Development Investment Group Co. Ltd., bringing together state-owned capital at the provincial, municipal and county levels.
From the outset, the company carried a policy mission. It was included in China’s 13th Five-Year Plan for major integrated-circuit production capacity, with one core goal in DRAM: break into a market dominated for decades by Samsung, SK Hynix and Micron.
At the time, China had three major memory projects. Yangtze Memory Technologies focused on NAND flash, ChangXin pursued self-developed DRAM, and Fujian Jinhua planned to work with UMC to develop niche DRAM products.
UMC and Jinhua signed a technology cooperation agreement. Jinhua would fund line construction and equipment purchases, while UMC would send a core technical team to jointly develop 32nm DRAM technology. Jinhua put up $300 million to buy R&D equipment and paid UMC another $400 million over time, with both sides sharing the resulting technology.
Chen became Jinhua’s general manager in February 2017. Media reports cited in the article say he barely hesitated when invited. At that stage, Jinhua did not even have a proper R&D building. Its temporary office was set up in an idle factory building near the industrial park, looking out onto an active construction site. Asked why he joined the UMC-mainland DRAM development effort, Chen said the sale of Rexchip to Micron had left a deep impression on him, and independent DRAM development had long been a personal goal.
Progress moved faster than expected. The fab was built in less than two years. Under the original plan, phase one would involve total investment of $5.3 billion, begin production in the third quarter of 2018, and bring in 60,000 12-inch wafers per month using a 32nm process. The longer-term target was a 20nm product, with four phases completed by 2025 and monthly capacity reaching 240,000 wafers.
Micron litigation, the entity list and a criminal case
The first legal blow landed in September 2017, when Micron sued UMC in Taiwan, alleging that employees who moved from Micron to UMC stole DRAM trade secrets and helped develop a 32nm DRAM process. In December that year, Micron also filed suit against Fujian Jinhua and UMC in federal court in California.
Jinhua responded with a countersuit, alleging that Micron products sold in China infringed its patents. The company eventually won that case.
The more damaging move came in late October 2018. The U.S. Commerce Department placed Fujian Jinhua on the export control entity list on national security grounds. The restriction did not stop with U.S. companies. Global suppliers using U.S. technology were also barred from supplying Jinhua. For a newly built line that depended on a global supply chain, the order effectively cut the line in half.
One month later, the U.S. Justice Department formally charged Fujian Jinhua, UMC and Chen, among others, turning what had been a civil dispute into a criminal case. The counts included conspiracy to commit economic espionage.
In October 2020, UMC reached a settlement with the Justice Department. It admitted to misappropriating one trade secret, agreed to pay a $60 million fine, and promised to cooperate with the U.S. government’s investigation of Jinhua during a three-year probation period. Two months later, a U.S. court issued a global arrest warrant for Chen.
Because neither mainland China nor Taiwan has an extradition treaty with the United States, Chen remained in China and continued leading R&D efforts.
The standoff started to ease at the end of 2023, when Micron announced a global settlement with Fujian Jinhua. Both sides withdrew lawsuits against each other worldwide and ended all litigation. Two months later, on Feb. 27, 2024, the San Francisco federal court ruled that prosecutors had failed to prove Jinhua stole Micron proprietary data, clearing the company of all charges.
That outcome did not erase the delay. The six-year case materially slowed Jinhua’s timeline. In January 2025, after his employment contract expired, Chen stepped down as general manager and moved into the role of technical adviser.
Why DRAM is one of the hardest chip businesses
To understand Jinhua’s position, the article argues, it helps to look at the structure of the DRAM business itself.
At its core, DRAM is a huge array of memory cells. Each cell contains a transistor and a capacitor. Transistors can keep shrinking, but capacitors are much harder to miniaturize. Once the capacitor gets too small, it struggles to hold charge steadily and becomes more vulnerable to leakage and interference from nearby components.
Manufacturers have had to adopt increasingly complex structures to improve performance, which makes DRAM expensive to build. According to the article, a leading-edge DRAM fab now typically requires $15 billion to $20 billion in investment, not counting the additional billions needed for lithography, etching and other tools.
DRAM is also a commodity product built around common standards. Chips made by one supplier can be inserted directly into another supplier’s system. That has made the business especially sensitive to cycles in price, scale and cost.
The industry’s history is full of reversals. Intel dominated in the 1970s, at one point reaching an 82.9% market share. In the 1980s, Japan’s NEC, Hitachi and Toshiba used lower prices to overtake U.S. producers, and by 1985 Japan’s share had climbed above 50%. Samsung entered DRAM in 1983 after acquiring Korea Semiconductor.
After 1990, the market moved into a downturn. While the sector absorbed losses and DRAM prices collapsed, Samsung borrowed aggressively and expanded capacity, using scale to drive costs below what rivals could support. By 1998, Korean companies had overtaken Japanese firms in market share. In 2001, Hyundai was spun off and renamed Hynix, while Micron kept growing through acquisitions.
The 2008 financial crisis brought another reset. Germany’s Qimonda and Japan’s Elpida both went bankrupt, marking the exit of Europe and much of Japan from the front line of DRAM. The market then consolidated into the three-player structure of Samsung, SK Hynix and Micron, with combined share above 95%.
The survivors stopped short of launching all-out price wars and shifted competition toward technical positioning. High Bandwidth Memory, or HBM, is a clear example. It depends not only on process scaling but also on advanced stacking and packaging, including 3D stacking and through-silicon vias, or TSV. In these areas, Samsung and SK Hynix have already built technical barriers.
China’s domestic backdrop was difficult as well. When Qimonda collapsed, local state investors in China tried to acquire all of its assets, but the effort never went through after years of internal and external complications. Around 2015, chips overtook crude oil to become China’s largest import category. At that point, memory-chip pricing and supply quotas were still largely dictated by Korean and U.S. suppliers, and domestic production was near zero.
The article says Micron had reason to pay close attention to Jinhua for two reasons. One was the arrival of senior engineers and technical staff at the company. The other was Micron’s exposure to the mainland Chinese market. In 2016, Micron posted $12.4 billion in revenue, with more than 40% coming from memory purchases in mainland China. DRAM and NAND flash accounted for more than 90% of its sales.
In the article’s telling, that dependence on one market set the commercial backdrop for Micron’s cross-border IP lawsuits and its push to constrain Jinhua through the entity-list process.
The comparison with ChangXin is also telling. ChangXin also recruited staff from Samsung in its early years, and some cases drew the attention of the Korean legal system. But it also paid heavily to acquire Qimonda’s patent portfolio and signed a patent licensing agreement with Rambus, securing rights to implement a large body of DRAM technology. Zhu Yiming later said in 2019 that ChangXin had modified part of its technology stack to fully remove U.S.-related technology.
Rebuilding under sanctions
Once Jinhua was put on the entity list, its challenge became stark: rebuild an operable DRAM line without access to U.S. equipment and materials.
The article links that effort to an earlier episode in Chen’s career. During the 2008 financial crisis, Chen was general manager of Rexchip in Taiwan. The company, working closely with Micron, had a technically advanced DRAM line, but wafer yield had fallen to just above 60%. Every wafer lost money, and management was debating whether to shut the line down and sell the equipment as scrap.
Chen led a technical team that stayed with the line for four months, re-optimizing production data and processes. Yield eventually climbed above 80%. A line that had been losing $3 million a month turned into one generating $2 million a month, according to the article. Inside the company, that story later fed the nickname “yield magician.”
The article cites market talk that around 70% of Jinhua’s original equipment relied on U.S. suppliers. After the sanctions hit, Chen’s team pushed to modify domestic tools and rebuild process logic on the ground. The work was slow and painful, but it significantly reduced the share of U.S. technology in the equipment base.
Jinhua’s own website features a page titled “Jinhua Milestones.” The article notes that what appears there is not the noise of the dispute but a record of product progress and revenue milestones, reflecting a slow recovery under intense pressure.
Even so, the company remains well behind peers founded in the same period. ChangXin reached volume production of 19nm DRAM in 2019 and posted 50.8 billion yuan in revenue in the first quarter of 2025, the article says. Yangtze Memory mass-produced 232-layer 3D NAND in 2022 and was pushing toward a 9% market share in 2025.
Jinhua, however, has not fallen out of the game. It focuses on niche DRAM products used in smart TVs, set-top boxes, printers, routers and industrial control equipment. These product categories often carry life cycles of five to 10 years, and once a supplier gets into the chain, the position can be relatively stable.
Its 12-inch fab is now producing about 40,000 wafers a month, with capacity slated to expand to 60,000 in 2026. The company holds 1,007 related patents, covering the full process from chip design to manufacturing. But Jinhua is still on the U.S. entity list, and the article says the Commerce Department tightened restrictions further at the end of 2024. Legal obstacles may have eased, but the technology blockade has not ended.
Why Jinjiang bet on chips
Jinjiang is better known for shoes, apparel and food than semiconductors.
The city is often treated as a symbol of China’s private-sector economy. The article says private businesses contribute more than 90% of local GDP, and roughly one in seven residents is a business owner. Men’s apparel brands account for about 25% of national output, sports shoes about 20% of global output, snack foods about 20% of national output, and exterior wall ceramics about 60% of national output. Brands such as Anta, Xtep, 361 Degrees, Erke, Panpan, Hengan, Lilang and Joeone all come from the city.
Against that backdrop, a move into memory chips looked unusually aggressive. Of the three major storage projects launched around the same period, Wuhan had a 10-year foundation in XMC, and Hefei committed resources at the provincial level. Jinjiang had almost no semiconductor base at all. The first-phase investment in Jinhua, 37 billion yuan, was close to several times the city’s fiscal revenue at the time.
Jinhua became the seed project for Jinjiang’s IC industry, which is why the city moved so hard and so fast at the beginning.
Support infrastructure followed. In November 2017, the Fujian provincial government approved the Quanzhou Semiconductor High-Tech Industrial Park as a provincial-level zone with “one zone, three parks.” The Jinjiang section would focus on integrated circuits, Nan’an on compound semiconductors, and Anxi on optoelectronics led by LED manufacturing.
That citywide commitment mattered during Jinhua’s darkest stretch. Fujian state capital has held an absolute controlling stake of more than 60% in the company. Through five frozen years, the article says, there was no capital withdrawal, no pullback in lending, no breakup of the team and no interruption in payroll. Provincial and municipal authorities continued supporting plant operations, utilities and industrial resources.
The article quotes a local government official describing Jinjiang’s role as “guide, pusher and service provider,” with a simple principle: show up when called, do what is promised, do not interfere when companies can manage on their own, and give them a push when they are struggling uphill.
On financing, Jinjiang has set up two industrial parent funds and 12 government-guided funds as part of a provincial fund-cluster strategy, creating two fund clusters with total scale above 45 billion yuan. In March 2025, it also launched a 500 million yuan talent and technology innovation fund, with an initial tranche of 50 million yuan, aimed at integrated circuits, nuclear technology applications and intelligent equipment.
Using Jinhua as an anchor, Jinjiang has brought in 52 integrated-circuit supply-chain projects with total investment above 100 billion yuan, according to the article, forming a full-chain cluster spanning chip design, manufacturing, packaging and testing, equipment, materials, supporting services and end applications.
For 2024, output value from the city’s IC companies above designated size was expected to exceed 10 billion yuan, up 51% year on year. For 2025, that figure is expected to top 14 billion yuan.
Over more than 20 years, Jinjiang’s GDP rose from 27.7 billion yuan in 2002 to 336.35 billion yuan in 2023. In the first three quarters of 2024, GDP reached 249.46 billion yuan, up 8.2% and ranking first in growth among the country’s top five county-level economies.
Still in the race
China’s memory industry no longer looks like it did eight years ago. The article says the global DRAM market reached $40.037 billion in the third quarter of 2025, a record high for any quarter. Morgan Stanley expects the AI-driven memory supercycle to continue for years and forecasts the global memory market will exceed $300 billion by 2027.
In that context, Jinhua’s planned 60,000-wafer monthly capacity and 2 billion yuan in annual revenue may look modest. But in the article’s framing, the company stands for something else: proof that a business once hit by severe restrictions has managed to stand back up.
The original piece was published by the WeChat account Touzhong and written by Zhang Xue.