[Jeon Byeong-seo, Director of the China Economic and Financial Research Institute; Reporter Lee Dae-il, SOYEON KIM ] If asked whether Changxin Memory (CXMT)’s IPO poses an immediate threat to SamsungElectronics and SK hynix, the answer is “not yet.” CXMT is filling the supply gap in general-purpose DRAM created as SamsungElectronics and SK hynix shifted their focus to High-Bandwidth Memory (HBM) and high-performance server DRAM. At present, CXMT is a complementary player rather than a competitor.

◇ CXMT’s Market Share Projected to Reach Double Digits by 2028

The issue lies beyond 2028. CXMT’s global DRAM market share is projected to expand from 8% in the first quarter of this year to 11% by 2028. The $8.5 billion (approximately 12.5 trillion won) raised through this initial public offering (IPO) will be invested in the transition to Generation 5 DRAM fine-pitch processes and the mass production of HBM3. The company plans to build new fabs in Shanghai and Beijing and establish a production cluster centered on Hefei. It has also unveiled a long-term blueprint to double its production capacity by 2030 and triple it by 2035.

The most underestimated scenario is the day when restrictions on extreme ultraviolet (EUV) lithography equipment from the Dutch company ASML are lifted. On that day, CXMT’s pace of catching up will be on an entirely different level from what it is now. Ten years ago, no one took BOE seriously, and no one feared CATL. We already know the outcome.

U.S. sanctions have, in fact, fueled CXMT’s growth. When the U.S. imposed sanctions on Huawei in 2019 and tightened export controls on semiconductor equipment from 2022 to 2025, China declared its commitment to semiconductor self-reliance and nurtured CXMT. CXMT is a product of that sense of urgency. U.S. sanctions failed to cripple China’s semiconductor industry; instead, they instilled a sense of urgency that China could not survive without building its own ecosystem. The company pre-purchased large quantities of equipment before the regulations took effect and accelerated the transition to domestically produced equipment in the cleaning and etching sectors. The result was a dramatic turnaround from a 19.2 billion yuan loss in 2023 to a net profit of 50–57 billion yuan (approximately 10–12 trillion won) in the first half of 2026.

The technology gap varies by product category. In the LPDDR5 sector—a low-power DRAM—CXMT has closed in on the global leaders. In DDR5, it lags behind South Korean companies by about one generation. In HBM, CXMT is at the HBM3 sample stage, while South Korea is mass-producing HBM3E, creating a gap of more than two years. However, we must not mistake this gap for a permanent one. We have already witnessed the pace at which companies like BOE in liquid crystal displays (LCDs) and CATL in batteries have caught up. There are reports that Apple is currently testing CXMT’s DRAM. The fact that Tencent, Alibaba, HP, and Dell have adopted CXMT’s memory indicates that this is no longer merely DRAM for the Chinese domestic market.

◇ HBM Technology: A Huge Gap… Urgent Need for South Korean Government Support

SamsungElectronics’ most urgent task is to regain its HBM market share. According to market research firm Counterpoint Research, based on first-quarter revenue this year, SK hynix leads the global HBM market with a 58% share, while SamsungElectronics ranks second with 21%. The moment this gap becomes entrenched, CXMT’s pursuit in the general-purpose DRAM market will become even more devastating. Without dominance in the high-value-added market, it will be difficult to withstand price competition in the general-purpose market.

SamsungElectronics and SK hynix are generating significant profits not only from HBM but also from standard DRAM. The greater a company’s reliance on standard DRAM, the greater the threat it faces. If CXMT’s low-price offensive gains momentum, a decline in profitability will be inevitable.

Accelerating their technology roadmap toward HBM4 and HBM5 is a key challenge for SamsungElectronics and SK hynix. By the time CXMT enters the HBM3 space, South Korea must already be in the next generation.

The South Korean government has two key roles to play. First, concentrated support at the national level is needed for research and development (R&D) into HBM and next-generation AI memory. The Chinese government, which is fostering the semiconductor industry as a national strategic sector, continues to provide massive support to CXMT. China has nurtured CXMT through national will. South Korea, however, relies solely on the private sector.

Second, South Korea must maintain and cooperate with U.S. export controls on semiconductor equipment to China at the alliance level. Whether EUV restrictions remain in place is the most critical external factor determining the pace at which CXMT catches up. It is the government’s responsibility to manage this variable through diplomatic channels.

CXMT is living proof of China’s semiconductor rise—having shed the stigma of being a subsidy-dependent state-owned enterprise through its performance and emerging as the undisputed fourth player in the global DRAM market. It remains to be seen how this company, which went public today, will transform the memory market in ten years. However, if we do not take it seriously now, we will regret it ten years from now. Ten years from now, the memory market will see a deepening “K-shaped” divide, with “HBM and AI memory” dominated by South Korea and the U.S., and “general-purpose DRAM” becoming China’s domain. South Korea’s only survival strategy is a race to establish a decisive lead in AI memory. Without a decisive lead in HBM, there is no future. China is waging a national-level battle. In South Korea, where only companies have been fighting so far, the government must now step onto the battlefield as well.