In 2016, two significant events unfolded in China's memory industry, bringing two inland cities over 300 kilometers apart together through a remarkable twist of fate. Few could have predicted that a decade later, they would evolve into industrial powerhouses envied by others.
That March, the National Memory Base, with a total investment of $24 billion, was officially launched in Wuhan's Optics Valley. The scene was bustling with excitement, and local media described it as "the largest single high-tech industrial investment project in our province since the founding of the nation." Four months later, Yangtze Memory Technologies Co., Ltd. (YMTC) was established, targeting the flash memory market and later becoming one of China's top two memory giants.
In the same year, Hefei saw the founding of ChangXin Memory Technologies (CXMT), the other memory giant focusing on DRAM. In May, Zhu Yiming, founder of GigaDevice, and leaders from the Hefei Economic and Technological Development Zone discussed the development plan for local memory projects. One of them had previously competed with Wuhan for the location of the National Memory Base, while the other, a semiconductor entrepreneur with ambitions to become "China's top memory company" from the start, had once planned to acquire Beijing Sicreat, a company specializing in automotive memory, for 6.5 billion yuan but ultimately failed.
The two "disappointed individuals" quickly found common ground. This year was dubbed the "inaugural year of China's self-built memory industry" in the country's memory sector. A decade later, both YMTC and CXMT are accelerating their IPO efforts. This year also marks the onset of a super cycle in the global memory market, with giants like Samsung, SK Hynix, and Micron Technology reaping substantial profits. Even SK Hynix's salaries made headlines.
Wuhan and Hefei, the two memory hubs, have thus captured public attention, with some saying they "bet and won." However, the memory industry has always been a brutal game. China's prolonged struggle to break through in memory spans far more than a decade and involves more than just two cities and two companies. What appears to be a triumphant underdog story is, in reality, the result of prolonged dormancy and arduous breakthroughs driven by urban resource allocation, industrial capital investment, and the gathering of top-tier talent. There are no shortcuts—only countless attempts, revenges, relays, and fluctuations across cities and individuals.
The origins of China's memory chips were not in Wuhan or Hefei but in Beijing and Shanghai, the two semiconductor strongholds. Rewind to 1975: the semiconductor research group at Peking University's Physics Department, using technology from China's first semiconductor device manufacturing plant, "Factory 109," developed the mainland's first 1KB DRAM, five years behind Intel and three years before Micron's founding.
During the same period, Shanghai's semiconductor industry advanced alongside Beijing. Shanghai No. 19 Radio Factory and Beijing's Dongguang Electronics Factory (Factory 878) were known as the "two giants" of China's IC industry. The trial-and-error efforts in laboratories based in Beijing and Shanghai planted the seeds for China's memory industry. That generation of researchers started from scratch, forging a path for China's memory chip breakthroughs in an era "without materials, blueprints, or experimental equipment."
However, they might not have anticipated how challenging the journey from research to industrialization would be for Beijing and Shanghai. Projects like 907 (Shaoxing) and 908 (Wuxi) encountered numerous issues, with strategies prioritizing speed and cost-effectiveness leading to prolonged construction cycles and near-stagnation in domestic memory chip R&D and equipment manufacturing. By the 1990s, China decided to trade market access for technology, making joint ventures with foreign memory giants a key focus.
Japanese memory giant NEC arrived, partnering with Shougang to establish Shougang NEC, which mass-produced 4Mb DRAM in 1994. Simultaneously, it collaborated with Huahong in Shanghai, leveraging the 909 Project. Huahong NEC built China's first 8-inch production line and successfully mass-produced memory. But as noted, memory is a brutal game. South Korean memory manufacturers seized the opportunity, betting on national destiny. By 1998, Korean companies had surpassed their Japanese counterparts in memory market share. NEC and Hitachi later merged their memory divisions to form Elpida in a bid for survival, only to face bankruptcy again.
Ultimately, Shougang returned to its steel business, and Huahong shifted to foundry services. However, the failed experiences of Beijing and Shanghai in R&D, factory construction, and joint ventures cultivated China's first generation of semiconductor manufacturing talent.
The memory industrialization banner was taken up by Wuhan, a city that fought for it in its own understated way. The stories of CXMT and YMTC's beginnings are inseparable from one company: SMIC. In 2000, the State Council's "Document No. 18" sparked the first wave of chip entrepreneurship. That same year, Richard Chang returned to China to found SMIC, and the Wuhan government began planning to develop integrated circuit-related industries.
Wuhan's aspirations were urgent and straightforward. As a central city relying heavily on traditional industries like steel, it desperately needed a new industrial engine for the next decade. Thus, Wuhan's leaders set their sights on the integrated circuit industry. "If we don't do this, Hubei and Wuhan will have no place on China's future semiconductor map," said Zou Xuecheng, director of the Ultra-Large Scale Integrated Circuit and System Research Center at Huazhong University of Science and Technology, in an interview with the Changjiang Daily.
Over two years, Wuhan approached over ten renowned domestic and international chip manufacturers for collaboration. A chance opportunity led to an invitation for SMIC executives to visit Wuhan. In 2006, Wuhan Xinxin Semiconductor Manufacturing Corp. (XMC) was established, funded by the Hubei Provincial Government, Wuhan Municipal Government, and Donghu High-tech Zone, with the goal of building the first 12-inch integrated circuit production line in central China. Initially, SMIC managed its operations entirely.
However, the good times didn't last. At the critical stage of XMC's production launch in 2008, the financial crisis hit, plunging the semiconductor industry into a downturn. Born into adversity, XMC was forced to rely on external support, leveraging China's manufacturing cost advantages to provide foundry services for American flash memory company Spansion. But Spansion later went bankrupt, leaving XMC in dire straits again.
SMIC, facing its own struggles, couldn't continue injecting funds into XMC. Semiconductor giants like TSMC and Micron saw an opportunity to acquire XMC at a low price. Yet, the Wuhan government persisted, deciding to develop XMC independently. In 2011, the two parties established a joint venture. In 2013, SMIC exited XMC, and the Wuhan East Lake Development Zone invited former SMIC COO Simon Yang to join XMC as CEO. Yang had over a decade of experience at Intel.
Under the new CEO and with support from Wuhan's "national team," XMC secured orders from key clients like GigaDevice. By 2014, the National Integrated Circuit Industry Investment Fund initiated a second major investment in China's semiconductor sector, and Wuhan decided to "gamble once more" in memory.
The National Memory Base was something Wuhan fought for. According to media reports at the time, upon taking office, then-Wuhan Mayor Wan Yong visited the East Lake High-tech Zone to research the National Memory Base project and immediately decided to submit a proposal at the National People's Congress. At the 2015 National People's Congress, Wan Yong pleaded for national support to establish the National Memory Base in Wuhan and visited relevant leaders at the Ministry of Industry and Information Technology.
Soon, news spread that the National Memory Base was selecting a location, with cities like Beijing and Hefei joining the competition. According to later government investment promotion personnel, in the nearly three years before the base landed in Wuhan, they almost followed leaders to Beijing weekly. Perhaps it was the relentless determination of Hubei people that ultimately led to Wuhan being chosen. Building on XMC, the National Integrated Circuit Industry Investment Fund, Tsinghua Unigroup, and the Hubei and Wuhan governments jointly established YMTC.
This was the beginning of YMTC—not a热血爽文 (thrilling underdog story), but a hard-won persistence.
While Wuhan finally gained the favor of the "national team," Hefei at the time resembled a spirited youth. Having invested in "panel leader" BOE, Hefei, known as the "gambling city," made a name for itself. The story of "investing to pick up bargains," which accounted for nearly 40% of its fiscal revenue that year, later became the津津乐道 "Hefei model."
Hefei practiced a mechanism with BOE: "using equity investment thinking for industrial introduction and investment banking methods for industrial cultivation"—significant equity investment to pull projects in, then exiting at the right time to reinvest the returns. According to statistics from New Fortune magazine, Hefei's state-owned capital achieved a paper profit of about 14 billion yuan from three investments in BOE, with a return rate exceeding 120%.
Investing in BOE not only brought profits to Hefei but also highlighted industrial shortcomings. To upgrade Hefei's local home appliance manufacturing industry, screens were not enough—chips were needed. Hefei recognized the growing demand for chips in smart home appliances. In 2020, Hefei's leaders used an idiom to describe the layout of emerging industries: "芯屏器合" (a play on words meaning "chips, screens, devices, and integration," homophonous with "心平气和" or "calm and composed").
Thus, for Hefei, memory chips were not a gamble but a顺势而为 (move with the trend). A semiconductor investor mentioned that around 2016, Hefei was already bringing in many semiconductor industry professionals as investment advisors, and government investment promotion personnel were actively承接 (taking on) projects溢出 (overflowing) from chip hubs like Shanghai and Beijing.
After losing the competition with Wuhan for the national-level project, Hefei didn't give itself a breather. In May of the same year, the Hefei 506 Project began酝酿 (taking shape)—the largest industrial project ever invested in by Anhui Provincial Investment Group Holding Co., Ltd., aimed at creating China's first, largest, and most technologically advanced self-developed DRAM manufacturing project. CXMT was thus born, operating in a "BOE-like model," with the Hefei government providing initial funding and the company responsible for market-oriented operations.
Unlike Wuhan's existing memory chip foundation, Hefei built its memory chip industry from the ground up. In the early stages, Zhu Yiming, founder of GigaDevice, approached former SMIC CEO David Wang. At the time, Wang was already 70 years old, known in the Chinese semiconductor circle as the "Applied Materials patent king." He had worked at Applied Materials for 25 years and was the highest-ranking Chinese technology executive in Silicon Valley.
According to Caixin reports, early employees of CXMT were drawn by David Wang. Under his rallying call, CXMT built its initial core team from Taiwanese chip companies like TSMC and United Microelectronics Corporation (UMC), also吸收 (absorbing) some local talent from XMC, SMIC, and SK Hynix's Wuxi factory.
This was the beginning of CXMT—not a热血爽文, but a顺势而为.
Let's revisit the残酷的江湖 (brutal arena) of memory. Renowned semiconductor investor Chen Qi mentioned in "The Memory War" that the memory industry plays a循环游戏 (cyclical game) of "investment-overcapacity-oversupply-consolidation and clearance-shortage-investment": "Under every grain lies the blood and tears of giants."
The memory industry also follows Moore's Law, with integration density doubling every 18 months, meaning performance doubles while unit prices halve. Over 60 years of development, every iteration of mainstream products and every technological evolution requires players to go all out. The industry has undergone two regional shifts—from the U.S. to Japan, and from Japan to South Korea—adhering to basic principles:制定扩产策略 (formulating capacity expansion strategies based on demand), leveraging economies of scale to reduce costs, daring to invest counter-cyclically in下一代工艺 (next-generation processes), and continuously improving technology and products to form a positive循环 (cycle).
Whether for Wuhan and YMTC or Hefei and CXMT, their rise was never overnight. Behind China's艰难突围 (arduous breakthrough) in memory are countless cities chasing, countless individuals seeking revenge, and countless failures. While Wuhan and Hefei stepped into the spotlight with memory, many are unaware that cities like Jinjiang, Xi'an, and Chengdu also had memory ambitions. In the same year YMTC and CXMT were founded, Jin Hua Integrated Circuit Co., Ltd. was also established, and they were even called the "three carriages" of China's memory chip industry.
However, less than two years after its founding, Jin Hua encountered狙击 (sniping) from U.S. patents and regulations. In 2017, memory giant Micron sued Jin Hua, accusing it and UMC of infringing on its DRAM trade secrets. It wasn't until 2023 that Micron and Jin Hua reached a global settlement, but the six-year dispute severely hindered Jin Hua's development.
It's not just about countless cities exploring; there are also countless individuals and countless failures. After YMTC's establishment in 2016, Charles Kao, known as the "Godfather of Taiwan Memory," served as its executive director and acting chairman. This veteran with over 30 years of experience in the semiconductor industry, after retiring from Nanya Technology, plunged into mainland China's memory arena.
His choice to折腾 (stir things up) again stemmed from a sense of humiliation. In 2010, when Kao's eldest son took a position at Samsung in South Korea, considering同行避嫌 (avoiding conflicts of interest in the same industry), he主动 (proactively) explained the situation to a Samsung Electronics vice president. To his surprise, the response was indifferent: "We don't even consider you (Nanya Technology) as competitors."
To争一口气 (prove a point), Kao returned to YMTC to start anew. He assisted YMTC in getting on track, stating in an interview: "I've been on the mainland for many years. The mainland is not weak; it's just that many people think it is." Another individual choosing to seek "revenge" in China was坂本幸雄 (Yukio Sakamoto), the "Godfather of Japanese Semiconductors," who came to Tsinghua Unigroup in his 70s.
In the 1990s, when Japan's memory industry was in困境 (trouble), Sakamoto was appointed to lead the restructured Elpida. However, after the financial crisis, in a price war with Samsung, Elpida ultimately went bankrupt and was acquired by Micron. Watching Japan's largest and last memory manufacturer fall, Sakamoto once said: "I don't want to end my life as a loser; I want to settle this myself."
On the other hand, Zhu Yiming's story seems somewhat悲壮 (tragic yet heroic). On July 16, 2018, the day CXMT announced正式投片试产 (official tape-out and trial production), GigaDevice announced that Zhu Yiming had resigned as the company's general manager, taking over the leadership role from David Wang. When starting anew, he made a军令状 (military pledge): until CXMT becomes profitable, he would not take a single yuan in salary or bonuses.
Under his leadership, CXMT acquired a large number of technology patents from Qimonda (formerly Infineon, Europe's memory leader) from Canadian intellectual property company WiLAN and hired several资深 (senior) technical talents from Qimonda as advisors, thereby strengthening the company's technological储备 (reserves). Zhu Yiming is also known for being低调务实 (low-key and pragmatic), rarely giving media interviews and emphasizing "埋头苦干" (keeping one's head down and working hard) within the company.
Looking through the list of Chinese memory companies, there are countless烂尾 (unfinished) projects and failures—some from炒作 (hype) during泡沫时期 (bubble periods), others from the bitterness of not being able to persevere. The "twin stars" of CXMT and YMTC, and the cities of Hefei and Wuhan, have传奇故事 (legendary stories) that are difficult to replicate. For memory companies, listing is not the终点 (end goal); for cities, the memory industry ecosystem still requires careful cultivation.
热血爽文 have scripts, but the memory industry does not. It relies on the perseverance of city leaders, the坚持 (persistence) of entrepreneurs, the投入 (dedication) of engineers, day after day of enduring, and even some难以言说的运气 (inexplicable luck). As Charles Kao said: "In making semiconductor products, it's basically not about who is smarter, but about how much money you have, how many people, how much R&D experimentation you can do, and how fast you can do it. These are what matter most."
There are no myths in China's memory industry, nor in Hefei and Wuhan. Attempting to复制 (replicate) them is not advisable. What's more crucial is recognizing the规律 (patterns): they respect talent, respect capital, respect the market, and translate this respect into long-termism.
Comments