China, Which Couldn't Even Come Close to Samsung and Hynix...: The Truth Behind the 'China Shock' Rocking Stock Markets

* This article was produced based on a video published on July 31, 2026.

A pitch-black July struck the South Korean stock market. The KOSPI plunged more than 22% over the month of July. This was the result even after a record-breaking rebound on the final day. Even though SK Hynix hit its daily upper limit for the first time in history on that day, its stock closed down nearly 60% from its June peak. Semiconductor stocks worldwide fluctuated together throughout July, but the drop in South Korea's two market leaders was particularly steep. It was clear that the decline went well beyond a routine correction for stocks that had surged so rapidly.

Several reasons are being cited. Chief among them was fear over the possibility that the oligopolistic status Korean memory semiconductors enjoy globally—that is, the fundamental strength of Korean semiconductors—could be shaken. This bombshell exploded on Monday of that week. China has begun producing immersion DUV equipment in-house, and CXMT (ChangXin Memory Technologies) rose to become the largest company by market capitalization in mainland China on its first day of listing. These two headlines combined to produce an explosive force that wiped out the final week of July for the Korean stock market. Are these bombshells from China merely blank rounds? Or do they represent a structural threat signifying that the glory days of Korean memory semiconductors' market dominance are coming to an end?

---

They would not even confirm whether there was online information accessible through restricted channels. Unless one was an insider who already knew the necessary extension number, it was difficult to engage further. Aishengna Electronic Technology Group is a Chinese state-owned enterprise—the very company reported by U.S. media to have begun mass-producing immersion DUV lithography equipment. What exactly is immersion DUV, and why did this news immediately weigh down the stock prices of Samsung Electronics and SK Hynix in the final week of July?

Until now, Chinese memory semiconductor companies could not come anywhere close to Samsung and SK Hynix because they were unable to acquire this equipment. But if China is resolving the biggest bottleneck in its pursuit of Korean semiconductors through homegrown localization, does that mean Samsung and SK Hynix's market dominance has only a few years left? That is how the news was interpreted.

In truth, Samsung and SK Hynix faced several other negative factors. However, above all, news that the unique value proposition of Samsung and SK Hynix could be threatened triggered a domino effect of plunging Korean semiconductor stocks. The market debut of ChangXin Memory Technologies was the next blow.


Has China Pierced the 'Sanctuary' of Core Chipmaking?
Lithography equipment is, simply put, a printer. The light emitted by this machine acts like a pen, drawing semiconductor circuits onto wafers. It is one of the most essential processes in semiconductor manufacturing. Worldwide, ASML of the Netherlands was the sole company capable of making the lithography equipment used in producing high-spec chips. The semiconductor supply chain consists of a few oligopolistic winners. Among them, the company that appeared completely irreplaceable was not Nvidia, TSMC, or Samsung and SK Hynix, but ASML.

The most advanced lithography equipment, extreme ultraviolet (EUV) systems, can be manufactured exclusively by ASML worldwide. The tier right below EUV is immersion deep ultraviolet (DUV), which China has reportedly begun mass-producing on its own. DUV has a coarser light tip than EUV. Therefore, it cannot draw intricate semiconductor circuits in a single pass like EUV does. However, "immersion" means immersing it in water. By placing ultra-pure water on the wafer, light is focused more effectively to increase resolution. It is a more advanced product than dry DUV, which did not use water, and is widely used alongside EUV in manufacturing premium chips. ASML produces 97% of the world's immersion DUV equipment as well. Japan's Nikon also manufactures it, but high-spec models remain ASML's exclusive domain.

That is why seven years ago, in 2019 during the first Trump administration, when the U.S. first determined that China's semiconductor ambitions and technological advances could no longer be left unchecked, it began blocking exports of ASML equipment to China. Without ASML machinery, Chinese semiconductors could not advance any further anyway—the U.S. had identified the exact bottleneck. The logic was: "ASML, U.S. patented technology goes into making your monopoly equipment. Even though you are a Dutch company, if the U.S. says do not sell, you cannot sell to China." That was how the blockades began.


Where Was It Hiding? China's Veiled Semiconductor Ambition
China has been unable to purchase a single EUV machine for seven years, and immersion DUV equipment has not entered China for three years either. Yet China remains ASML's single largest customer, accounting for one-third of its revenue. By running lower-spec DUV equipment that China finds suboptimal, along with legacy equipment secured before U.S. export controls took effect, China is known to be producing 7nm to 5nm class chips.

Simply put, in semiconductors, smaller numbers are better. Compared to South Korean chipmakers, which are pushing toward mass production at 2nm, Chinese semiconductor technology still lags by about two generations. Furthermore, while EUV draws semiconductor circuits precisely in a single pass, DUV requires multiple passes to achieve the desired pattern, resulting in higher defect rates. Even at the 7nm class, China's yield and productivity fall well short of South Korea's.

Even so, backed by massive support from the Chinese government, China's semiconductor industry reached a milestone just as news emerged that Aishengna had begun mass-producing immersion DUV inside China. Although established only three years ago, it did not appear out of thin air. It is a company created under a rebrand to hide from the U.S. so that Washington could not quickly track their pace of development. This newly established state-owned enterprise, inheriting the know-how of Chinese companies that had been researching related equipment, succeeded in mass-producing semi-advanced manufacturing equipment once exclusive to ASML. Aishengna is reportedly aiming to produce 5 units this year.
Can 5 units a year really be called mass production? For this specific equipment, it can. ASML, which holds a global monopoly on these semiconductor circuit printers, produced a total of just 327 units last year, combining EUV and DUV machines. An EUV machine costs over 500 billion won (approx. $365 million) per unit, while an immersion DUV machine costs around 150 billion won. And China plans to produce 5 of these this year and 20 next year.

Of course, this alone cannot power even a quarter of a semiconductor fabrication plant. But what happens after that? By all accounts, Aishengna's immersion DUV equipment clearly falls short of ASML's products that cost over 150 billion won. Its technical capability is reportedly at a level where it produces 7nm-class advanced chips at yields far lower than South Korea's—a trial-and-error approach where, regardless of defective output, the Chinese government continues to fund multi-patterning efforts to see if usable 7nm chips can eventually be made.

It is nowhere near a level that could threaten ASML's standing—unlike Japan's Nikon, which currently holds about 3% of the immersion DUV market. Manufacturing a handful of these units in China will not immediately crack the fundamentals of Samsung and SK Hynix, nor shatter their position as absolute leaders in the AI memory ecosystem or their share in the high-end commodity memory market.

However, the tremendous expectations reflected in the recent surge in Samsung and SK Hynix's stock prices were not just for this year. If China can roll out a low-spec immersion DUV machine in 2026, what will it be capable of by 2030? The problem is that no one knows the answer yet. This uncertainty is heightened because China works hard to keep the full picture of its semiconductor manufacturing capabilities under wraps. As question marks grew over whether China would trail by decades while following South Korea or close the gap within 10 years, combined with other headwinds, momentum built to sell off Samsung and SK Hynix shares.


'Mainland Market Cap No. 1' CXMT: Has It Already Infiltrated Samsung and Hynix's 'Wallets'?

Enlarge image

On top of this, ChangXin Memory Technologies (CXMT) listed on mainland Chinese exchanges on Monday of that week. CXMT already held a substantial presence, but was this the start of a full-throttle sprint? Was it a mere coincidence that news of China's localization of immersion DUV leaked beyond China's borders on the very day CXMT went public? Above all, CXMT shook the core expectation that drove enthusiasm for Samsung and SK Hynix this year: the belief that memory chip demand would break free from traditional industry cycles to remain persistently hot, and that memory chips would remain premium products obtainable only through costly, long-term contracts.

China remains the largest destination for South Korea's semiconductor exports. Based on revenue, 30% of South Korean chip exports still go to mainland China, and that figure reaches 45% when including Hong Kong. What about exports to Taiwan, which include cutting-edge High Bandwidth Memory (HBM) for AI? Even in the first half of this year, Taiwan accounted for only about 14% of total semiconductor exports. This means that even though Samsung and SK Hynix sell zero cutting-edge AI memory chips to China, they are still making money in China.

The surge in Samsung and SK Hynix stock prices in the first half of this year was not driven solely by AI memory. As Samsung and SK Hynix focused on AI memory such as HBM, reducing their capacity to produce legacy commodity memory, the world clamored for commodity memory chips. Prices for identical legacy DRAM and NAND flash skyrocketed 10 to 12 times in just one year, sending operating profit margins at Samsung Electronics and SK Hynix to unfathomable levels for manufacturing companies, and stock prices surged alongside rosy forecasts. But what happens if Chinese companies like CXMT and YMTC produce high volumes of commodity memory comparable to Samsung and SK Hynix's products, even without reaching HBM?
The AI boom brought immense opportunities to Chinese memory semiconductor companies. They rapidly moved into the vacuum left by Samsung and SK Hynix, which lacked the capacity to supply commodity memory to existing clients as before. Meanwhile, the Chinese government is urging domestic companies to use products from CXMT and YMTC, even if their quality is somewhat inferior.

Even U.S. companies are eager to increase their use of Chinese semiconductors whenever the opportunity arises. You may have heard that Apple has been lobbying the U.S. government intensely lately, effectively saying: "We can no longer afford chips from Samsung, SK Hynix, and Micron because they are too expensive; let us buy memory products from China's CXMT and YMTC too." Early last year, CXMT held only a 3% share of the global DRAM market, but that figure surged to 8% in just one year. There are even reports that the company aims to surpass third-place U.S. chipmaker Micron, which trails Samsung and SK Hynix, by 2030.

As soon as CXMT listed on the STAR Market—often dubbed China's Nasdaq—its stock price jumped nearly fivefold from its IPO price. It immediately rose to the top spot in market capitalization on mainland Chinese stock exchanges. While the initial public offering price was set very low and only a tiny fraction of total shares—just 6%—was floated during the listing, similar to Elon Musk's SpaceX, creating scarcity value, the company nevertheless plans to pour a significant portion of the massive funds raised into factory expansions. If CXMT expands supply while venturing into relatively advanced memory products like DDR5, which it still manufactures less effectively than South Korea, what will happen in a few years to Samsung and SK Hynix, which have achieved high operating margins through the sustained price surge of commodity memory while still generating revenue in China? This poses the second question mark regarding Samsung and SK Hynix's fundamentals.


Is the 'China Shock' to Korean Semiconductors Real?
In fact, looking at SK Hynix's second-quarter earnings and Samsung Electronics' conference call, there is no evidence that the fundamentals of these two companies are deteriorating. While SK Hynix posted extraordinarily strong performance, its revenue was indeed lower than the market's consensus expectations. Paradoxically, however, this was because SK Hynix has been signing many long-term contracts, offering prices slightly lower than the soaring spot market prices for semiconductors. When negotiating a five-year contract, prices are naturally discounted compared to immediate spot rates. This indicates that the memory semiconductor industry is transitioning away from being a purely cyclical business. As for China's AI model Kimi 3, unveiled in mid-July to the shock of the world, while OpenAI and Anthropic might be alarmed, the prevailing view is that it will actually boost demand for AI memory.

The reason Samsung Electronics and SK Hynix stock prices triggered sidecars almost daily throughout July was that single-stock 2x leveraged products, half of which were filled with semiconductor shares, rocked the Korean market. Financial authorities are belatedly introducing complementary measures like cap regulations. Moreover, even if the U.S.—which froze benchmark interest rates on July 30, Korean time—were to raise rates in September, and even if current interest rate anxieties persist, it seems clear that U.S. Big Tech companies cannot stop the AI race. Questions over how long these tech giants can continue competing as cash grows tight were one of the main factors dragging down semiconductor stocks in July, but there is still no signal that this race will halt anytime soon.

Yet the presence of Aishengna and CXMT, emerging from China early in the last week of July, served as a stark reminder that South Korea's semiconductor fundamentals are being pursued. The pursuer is still far behind. They are chasing, but a considerable gap remains. However, how fast are they actually moving? The fact that no one knows, and that question marks are growing, combined with other negative catalysts, marked the climax of the July crash in South Korean semiconductor stocks.
※ Please note: This article was translated by AI and may contain errors.