Chinas NAND Gamble Threatens Korean Duopoly

The Gap Opening Up

In the second quarter of this year, a Chinese company called Yangtze Memory Technologies Corp., or YMTC, overtook Kioxia in NAND flash shipments. It did not sound like a seismic event. But look closer at the numbers, and the implication is harder to dismiss.

YMTC is now racing toward number one in a market long dominated by two Korean firms. Samsung Electronics and SK Hynix have held the top two spots for over a decade. Their combined share has been close to two-thirds of global NAND revenue. That dominance is starting to erode—not from a single breakthrough, but from a gap that YMTC is widening every quarter.

The scale of the gap is stark. According to data firm Omdia, YMTC’s NAND wafer production will jump from 2.01 million this year to nearly 2.5 million next year—an increase of roughly 486,000 wafers. Samsung will add about 75,000. SK Hynix, including its Solidigm subsidiary, will add roughly the same. In other words, YMTC alone is expanding faster than both Korean giants combined.

Most of YMTC’s growth comes from its Wuhan Fab 3, which is expected to begin initial operations before year-end and ramp from about 90,000 wafers this year to 575,000 next year. That single facility will account for the vast majority of YMTC’s production increase. The plant is being built with state-backed financing and local government support, operating under a timeline that Korean firms simply cannot match.

Why the Koreans Are Pausing

Samsung and SK Hynix have not chosen to ignore NAND. They have made a calculated decision to redirect investment toward high-bandwidth memory, or HBM, which is essential for artificial intelligence chips. HBM commands far higher margins per wafer than NAND. The math is compelling: a single HBM stack can sell for ten times the value of a comparable NAND wafer.

This pivot has worked so far. HBM demand from AI chipmakers like NVIDIA and AMD has been insatiable. Both Korean firms have secured long-term supply contracts that lock in revenue through 2028. But the cost of that strategy is becoming visible. Samsung’s Pyeongtaek campus P5 facility—its newest and most advanced memory site—will not begin NAND production until 2027 at the earliest, with full capacity coming online between 2027 and 2030. SK Hynix confirmed construction investment at its Cheongju M17 site last month, with groundbreaking slated for February and the first cleanroom opening by December 2028.

In the meantime, YMTC is building faster. And it has a funding mechanism that neither Korean firm needs to worry about: an initial public offering that could raise tens of billions of dollars and be deployed directly into capacity expansion. Industry reports suggest YMTC’s IPO target—already set for as early as this fourth quarter—has internally projected goals of reaching global NAND market leadership by the end of next year.

That is not a vague ambition. It is a public statement of intent from a company backed by the Chinese state, operating in a sector Beijing has identified as strategically vital.

The Price Question

Here is where the situation gets complicated for Samsung and SK Hynix. YMTC’s current product mix skews heavily toward consumer-grade NAND—the kind used in smartphones, laptops, and storage drives. Its share of enterprise SSDs, the high-margin segment that has historically given Korean firms their profit cushion, remains relatively small. That is why YMTC ranks fifth globally by revenue despite its third-place position by shipment volume.

But NAND prices rose sharply in 2024 and 2025, driven by supply discipline and strong demand from AI infrastructure builds. As YMTC expands and gains access to more advanced process nodes, it will likely move up the product ladder. Even a modest shift toward enterprise SSDs would compress margins across the entire market.

TrendForce has already flagged the risk. Its latest outlook notes that NAND will face increasing downward price pressure from late 2027, as new capacity from China and elsewhere comes online just as consumer electronics demand begins to soften. Unlike HBM, where supply has been tightly constrained by a narrow group of qualified manufacturers, NAND has many players and no real bottleneck.

If prices fall while Korean firms sit on limited new capacity, the revenue gap between their DRAM/HBM businesses and their lagging NAND operations will widen further. That is a structural problem, not a cyclical one.

The Geopolitical Layer

This is not simply a commercial competition. China’s push into memory chips is part of a coordinated industrial policy that has been in motion for years. YMTC was established in 2016 with direct support from the state-backed China Integrated Circuit Industry Investment Fund, commonly known as the Big Fund. Despite U.S. export restrictions that have limited YMTC’s access to advanced lithography equipment, the company has developed its own crystal-elastic storage architecture—a 3D NAND design that bypasses some traditional process requirements.

The International Trade Commission has repeatedly flagged YMTC as a national security concern. Several U.S. legislators have called for restrictions on YMTC’s IPO proceeds being used for further expansion. Yet the IPO itself is proceeding, and the company appears to have enough domestic capital markets backing to absorb those pressures—at least for now.

For Samsung and SK Hynix, the geopolitical dimension adds urgency. Both companies depend heavily on U.S. technology and equipment, and any escalation in export controls could constrain their own capacity expansion timelines even further. Meanwhile, Chinese state subsidies allow YMTC to sustain losses that would be unacceptable to a publicly traded Korean firm.

What Comes Next

The next 18 months will be decisive. YMTC’s Wuhan Fab 3 ramp, its IPO proceeds, and the pace at which it moves into enterprise storage will determine whether its current third-place position becomes a serious challenge for the top spot. Samsung and SK Hynix, meanwhile, face a window between now and 2027–2028 where their NAND capacity remains essentially flat.

If NAND prices hold, the Korean firms can weather the gap. If they fall—as TrendForce and others predict—the impact will be felt most acutely by Samsung, which carries a larger absolute NAND operation than SK Hynix. SK Hynix’s smaller NAND footprint gives it more room to absorb margin pressure while it prioritizes HBM.

But this is not a story with a simple winner and loser. It is a story about a market that was stable for a decade suddenly finding its equilibrium shattered by a competitor with different economics, different timelines, and different backers. The Korean firms are not wrong to prioritize HBM. The question is whether they will look up in time to see what is happening in the segment they are leaving behind.