DRAM Giants Get Rich as China Builds a Threat

• Micron, Samsung and SK hynix are collecting extraordinary profits from soaring DRAM prices while customers face the consequences of a severe memory shortage.
• China’s CXMT is advancing DDR5, LPDDR5X and next-generation DRAM technology despite U.S. restrictions on advanced semiconductor manufacturing equipment.
• CXMT’s projected manufacturing capacity could increase 83% between 2025 and 2028, potentially threatening the pricing power that has made the established DRAM manufacturers so profitable.

The DRAM Giants Are Making a Killing From the Memory Shortage

Micron Technology (MU), Samsung Electronics and SK hynix (SKHY) are making a fortune from the global memory shortage, and their customers are paying for it. Artificial intelligence infrastructure investment has driven extraordinary demand for high-bandwidth memory (HBM), server DRAM and other advanced memory products, while constrained supplies have sent prices soaring. Wall Street is celebrating the resulting profits, but investors should be asking whether the industry’s three dominant manufacturers are becoming dangerously comfortable with a market in which scarcity generates exceptional returns.

Micron’s financial results expose just how lucrative the memory shortage has become. In fiscal 2026, the company generated $133.19 billion in revenue, up 256% from $37.38 billion in fiscal 2025. Its GAAP gross margin more than doubled from 39.8% to 80.7%, while net income surged from $8.54 billion to $84.97 billion. Most revealing, Micron reported that DRAM average selling prices increased approximately 180% year over year, compared with bit shipment growth in only the mid-20% range. The enormous expansion in profitability was driven primarily by customers paying dramatically higher prices, not a comparable increase in memory production.

The three manufacturers have continued spending heavily on new technology and production facilities, but their investments have increasingly favored the most profitable products, particularly HBM and advanced server DRAM. The strategy is rational from the perspective of maximizing shareholder returns, but its consequences extend beyond the companies’ income statements. Conventional DRAM supply remains constrained, customers face sharply higher prices, and the manufacturers benefit from a market in which additional production has not kept pace with demand.

Meanwhile, China’s ChangXin Memory Technologies (CXMT) is pursuing a strategy that could eventually undermine the very scarcity that has made the established manufacturers so profitable. Despite restrictions on advanced semiconductor manufacturing equipment and its inability to acquire EUV lithography systems, CXMT has continued developing more sophisticated DRAM technologies while preparing a substantial expansion in manufacturing capacity.

The irony is difficult to ignore. Micron, Samsung and SK hynix are collecting the rewards of today’s shortage while China is investing to make sure those rewards do not last forever.

The DRAM Giants Protect Their Margins While CXMT Builds Capacity

The memory industry has a long history of manufacturers expanding production too aggressively, creating oversupply and watching prices collapse. Following the severe downturn in 2022 and 2023, Micron, Samsung and SK hynix reduced production or moderated certain investments to restore profitability. That response was understandable, but the industry’s subsequent emphasis on protecting margins has become particularly lucrative as AI infrastructure spending accelerates.

HBM has become central to this strategy. Used alongside advanced AI processors, HBM commands premium prices and requires sophisticated manufacturing and packaging. The established manufacturers have increasingly directed capital and production resources toward these products, seeking the highest possible returns from an industry experiencing unprecedented demand.

There is nothing inherently improper about maximizing profits. The problem is that the three manufacturers may be confusing their ability to profit from scarcity with an enduring competitive advantage. The more attractive DRAM prices become, the stronger the incentive for China to develop its own alternative suppliers.

CXMT is responding to that incentive with manufacturing investment. Unlike the established producers, whose capital allocation decisions are closely tied to near-term financial returns, CXMT operates within China’s broader objective of reducing dependence on foreign semiconductor suppliers. Its investments therefore reflect strategic considerations that extend beyond the profitability of any individual memory cycle.

According to Table 1, CXMT’s estimated installed DRAM wafer capacity is projected to increase from 300,000 wafers per month at the end of 2025 to 550,000 by the end of 2028, representing growth of 83.3%. The expansion includes additional capacity at Hefei and Beijing, along with a new manufacturing contribution from Shanghai beginning in 2027.

The projected expansion is significant because additional wafer capacity is only one component of CXMT’s potential growth in memory production. Improvements in DRAM cell density can increase the number of memory bits manufactured on each wafer, while higher yields can increase the number of usable chips. If CXMT successfully combines capacity expansion with technological improvements, its growth in marketable memory output could exceed the growth in installed wafer capacity.

These forecasts are not guarantees. Equipment restrictions, process complexity, manufacturing yields and construction schedules could delay expansion. Nevertheless, the projections demonstrate the scale of China’s ambitions in DRAM manufacturing. While the established manufacturers enjoy extraordinary margins, CXMT is developing the manufacturing infrastructure needed to become a more consequential competitor.

The strategic danger is that the current shortage is making China’s investment case more compelling. High DRAM prices increase the economic value of domestic production, while dependence on foreign suppliers reinforces Beijing’s determination to develop alternatives. The established manufacturers may be maximizing today’s profits while strengthening the incentives for their future competition.

CXMT Is Challenging the Technology Gap, Not Just Adding Cheap Capacity

The established DRAM manufacturers could dismiss CXMT as a producer of older, less sophisticated memory products. That argument is becoming increasingly difficult to sustain.

CXMT has advanced beyond DDR4 into DDR5 and LPDDR5X, while pursuing next-generation mobile memory, advanced server DRAM and HBM technologies. These developments matter because the company’s ambitions now extend into product categories that have historically supported the established manufacturers’ technological advantages and premium pricing.

Industry estimates of CXMT’s 2026 product mix indicate approximately 25% DDR4, 25% DDR5 and 50% LPDDR5X. Although these are estimates rather than confirmed shipment shares, they suggest that CXMT is no longer concentrating primarily on legacy memory. The company is positioning itself to compete in increasingly sophisticated applications.

Its proposed 4F² DRAM architecture represents a potentially important technological development. Conventional DRAM memory cells commonly use a 6F² design, where F represents the minimum feature dimension. At the same feature dimension, reducing the memory-cell footprint from 6F² to 4F² decreases theoretical cell area by one-third, potentially increasing memory-cell density by 50%.

The distinction between theoretical density and commercial manufacturing performance is important. CXMT has not demonstrated that its proposed 4F² architecture can achieve competitive production yields, reliability or costs at scale. Samsung and SK hynix are also developing advanced memory-cell architectures, and the established manufacturers retain considerable manufacturing advantages.

Nevertheless, CXMT is attacking the technological problem from a different direction. Rather than relying entirely on smaller lithographic dimensions, it is pursuing architectural changes that can increase memory capacity within a given chip area. That approach is especially significant for a manufacturer operating without access to EUV lithography.

The lesson for investors is that equipment restrictions do not necessarily prevent technological development. They can increase manufacturing complexity and expense, but they do not eliminate the possibility of alternative process solutions.

SMIC Already Exposed the Weakness in the Sanctions Argument

The semiconductor industry should have learned this lesson several years ago.

In May 2022, I published a Seeking Alpha article examining Semiconductor Manufacturing International Corporation’s (SMIC) progress toward 7nm-class logic manufacturing without EUV lithography. At the time, many observers assumed that restricting China’s access to EUV systems would prevent its semiconductor manufacturers from advancing into more sophisticated process technologies.

SMIC demonstrated that the assumption was flawed. By using deep ultraviolet lithography and multiple patterning, the company achieved 7nm-class logic manufacturing despite the absence of EUV equipment. The approach involved additional manufacturing complexity and disadvantages in efficiency and cost, but the technological barrier was not absolute.

CXMT is pursuing a different objective. SMIC’s logic technology focuses on increasing transistor density, which can improve computational performance and power efficiency. CXMT’s DRAM technology focuses on increasing memory-bit density and the capacity of individual memory chips.

The common lesson is that Chinese semiconductor manufacturers are finding ways to advance technologically despite restrictions on their access to the most sophisticated foreign equipment.

The established DRAM manufacturers should not assume that today’s technological advantages are protected indefinitely by export restrictions. SMIC demonstrated the limitations of that assumption in logic manufacturing, and CXMT’s progress provides another warning in memory.

The Real Threat Is the Collapse of the Scarcity Premium

CXMT does not need to overtake Micron, Samsung or SK hynix technologically to damage their profitability. It does not even need to become a leading global supplier immediately. It only needs to add enough commercially competitive DRAM capacity to begin changing the balance between supply and demand.

China is a major consumer of memory semiconductors. Every additional chip produced domestically can potentially displace imported DRAM, reducing the market available to established foreign suppliers. Even if CXMT concentrates initially on Chinese customers, greater domestic production could force competing manufacturers to seek additional sales elsewhere, increasing pricing pressure.

That prospect becomes particularly important when the current shortage eventually eases. The established manufacturers are benefiting from a market in which customers have limited alternatives and demand has outpaced available supply. Those conditions have produced extraordinary profits, but they are not necessarily permanent.

The semiconductor industry’s history repeatedly demonstrates that exceptional margins attract additional investment and eventually encourage greater supply. What makes the present situation different is that the potential new competitor is supported by a national industrial strategy that places technological independence alongside commercial profitability.

For Micron, Samsung and SK hynix, this creates a competitive threat that cannot be dismissed simply because CXMT currently trails them in advanced manufacturing yields or HBM production. Those advantages matter, but they do not eliminate the consequences of a fourth manufacturer expanding its capacity and improving its products.

The danger is that the established manufacturers are enjoying the financial rewards of a shortage while underestimating how aggressively China is preparing for the next phase of the memory market.

Investor Takeaway

Micron, Samsung and SK hynix have every reason to celebrate their extraordinary financial performance, but investors should question whether the companies’ current profitability is concealing a developing competitive problem. The three manufacturers have benefited from sharply higher DRAM prices while directing enormous resources toward premium memory products. Their strategy has rewarded shareholders, but it has also coincided with technological progress and manufacturing expansion by a Chinese competitor whose objectives extend beyond immediate profitability.

CXMT’s advancement into DDR5, LPDDR5X and higher-density DRAM architectures demonstrates that U.S. equipment restrictions have not prevented continued technological development. Its projected 83% capacity expansion between 2025 and 2028 could make the company a more significant participant in the global memory market, particularly if its process improvements translate into higher yields and greater memory-bit output.

The greatest danger to the established manufacturers may not be CXMT’s latest technological announcement. It may be their own complacency. Micron, Samsung and SK hynix have become exceptionally profitable in a market where customers have limited alternatives, but those same extraordinary prices are strengthening China’s determination to develop alternative sources of advanced memory.

The DRAM giants are making a killing from today’s memory shortage. They may discover that their greatest mistake was giving China every reason to end it.

Contact [email protected] for any questions or corrections.