On August 19, Hanwha Securities estimated that China's ChangXin Memory Technologies (CXMT) will account for 11.3% of the world's net increase in DRAM bit supply in 2026. Even so, the supply-demand fulfillment rate including CXMT remains at negative 9.4%, falling short of demand. This figure does not support the view that China's DRAM production expansion marks the start of a global supply glut.
According to the same estimate, CXMT's contribution rate falls to 6.9% in 2027. However, this is not CXMT's market share or its own growth rate. It represents the proportion of newly added global DRAM bit supply that CXMT accounts for. Because production increases at Samsung, SK hynix, and Micron expand the denominator, CXMT's relative contribution appears smaller.
What matters is whether China's production expansion will exceed demand growth and supply constraints enough to create a surplus. Overlaying Hanwha's forecast with CXMT's capital expenditure plans and the supply constraints cited by Micron, the DRAM market from 2026 to 2027 appears to be in a phase where "increased production eases shortages" rather than eliminating them.
An 11.3% Contribution to Net Growth vs. a 6.7% Market Share
According to Hanwha's estimates, global DRAM bit supply growth will reach 22.4% in 2026 and 24.1% in 2027. Against this annual net increase, CXMT's contribution rate is 11.3% in 2026 and 6.9% in 2027. The denominator here is the net increase in global supply as a whole—not CXMT's shipment volume, nor its revenue-based market share.
For the same year of 2026, Hanwha places CXMT's share of global DRAM bit supply at 6.7%. While 11.3% and 6.7% may appear to be similar figures, the former refers to the annual increment, while the latter refers to total supply. Conflating the two would lead to different—and incorrect—conclusions, such as CXMT holding an 11.3% share of the global market or its supply growing by 11.3%.
CXMT's own IPO prospectus presents yet another denominator and time frame. Based on Omdia data, the document states that CXMT's DRAM revenue share in Q4 2025 was 7.67%, ranking it fourth globally and first in China. This is a quarterly revenue share figure, and it cannot be directly compared to the 6.7% bit supply share projected for 2026.
Even when discussing the same DRAM market, these figures measure different things. Without distinguishing between bit supply, revenue, and annual net contribution to growth, one risks overstating or understating the impact of Chinese manufacturers' production capacity on market prices and supply volumes.
How the Shortage Gap Changes With and Without CXMT
Hanwha's supply-demand fulfillment rate, including CXMT, stands at negative 9.4% for 2026 and negative 5.6% for 2027. A negative value indicates that supply falls short of demand. Excluding CXMT, the estimates are negative 15.5% for 2026 and negative 12.0% for 2027.
Comparing the two estimates, CXMT's supply narrows the shortage gap by 6.1 percentage points in 2026 and 6.4 percentage points in 2027—not a small change. At the same time, the fulfillment rate remains negative in both years even after accounting for the increased production, meaning the forecast does not project supply exceeding demand.
| Year | Fulfillment Rate (Including CXMT) | Fulfillment Rate (Excluding CXMT) | Gap Narrowed by CXMT |
|---|---|---|---|
| 2026 | -9.4% | -15.5% | 6.1 points |
| 2027 | -5.6% | -12.0% | 6.4 points |
In its public summary, Hanwha projects that global DRAM shortages will persist through 2027 even after CXMT's production increase, as new demand from Chinese servers absorbs the additional supply. An increase in supply and a resulting surplus in the global market are two separate outcomes. What the supply-demand model shows is that while CXMT changes the magnitude of the shortage, it does not reverse the overall direction of the supply-demand balance.
The Gap Between Wafer Capacity and Effective Bit Output
DRAM supply is not determined solely by the number of wafers fed into a factory. The number of effective bits that can actually reach the market from the same wafer varies depending on process generation, die density, yield, and product mix. This is why CXMT's capacity expansion cannot be translated into global bit supply on a simple one-to-one ratio with its physical scale.
HBM further complicates this calculation. Because HBM involves stacking and binning, it consumes more wafer capacity than conventional DRAM. In its earnings materials dated March 18, 2026, Micron cited HBM's high trade-off ratio as one of the supply constraints. The company also pointed to cleanroom and construction lead times, as well as the slowing pace of bit-per-wafer improvements during node transitions.
In the same materials, Micron forecasts that 2026 DRAM bit demand will be constrained by supply limitations, with tight supply-demand conditions persisting beyond 2026. This is Micron's own outlook as a market participant, and it does not directly verify Hanwha's supply-demand figures. Still, the premise that expanding capacity does not immediately allow supply to overtake demand provides useful context for interpreting Hanwha's forecast.
CXMT's own prospectus explains that ramping up a new production line typically takes several quarters, and that new factories can take years to move from capital expenditure to actual supply output. The same document states that, since the latter half of 2025, DRAM shortages and rising prices have persisted against a backdrop of computing demand and capacity allocation decisions by major manufacturers. There is a gap—in both time and manufacturing conditions—between announcements of mass-production capacity and actual bit supply delivered to the market.
Domestic Chinese Demand Absorbing New Supply
CXMT supplies DRAM wafers, chips, and modules, and has expanded its product lineup from DDR4 and LPDDR4X to DDR5 and LPDDR5/5X. Its prospectus lists partnerships with companies such as Alibaba Cloud, ByteDance, and Tencent. Having a domestic customer base spanning servers, smartphones, and PCs means that increased supply will not necessarily flow directly into overseas markets.
The scale of CXMT's business is already substantial. The company posted revenue of 61.799 billion yuan in 2025, with a compound annual growth rate of 160.78% from 2023 to 2025. TrendForce also projects that, driven by capacity expansion and yield improvements, CXMT will record the largest year-over-year growth in bit output worldwide in 2026. Where a company that already ranked fourth globally in DRAM revenue share in Q4 2025 directs its expanded output will matter just as much as the global supply volume itself.
Demand is also increasingly concentrated in servers. According to Omdia data cited in the prospectus, servers accounted for approximately 50% of DRAM demand in 2025, followed by mobile at approximately 28%, PCs at approximately 13%, and smart vehicles at approximately 2%. The same document cites an Omdia forecast that the server share will rise to approximately 71% by 2030. This demand structure underlies Hanwha's view that new server-related demand in China will absorb the additional supply from increased production.
CXMT plans to raise a total of 29.5 billion yuan through its offering, allocating 7.5 billion yuan to technical upgrades of memory wafer mass-production lines, 13 billion yuan to advancing DRAM technology, and 9 billion yuan to advanced DRAM research and development. While the scale of investment is substantial, TrendForce notes that equipment constraints exist for leading-edge nodes, and that new factories will initially focus primarily on DDR5. When assessing supply plans, tracking the ramp-up of mass-production lines and product mix offers a clearer read on effective global bit supply than investment figures alone.
Hanwha's estimate that CXMT's net contribution rate will fall to 6.9% in 2027 does not mean that the expansion of Chinese manufacturers is coming to a halt. Rather, it reflects how much the supply shortfall relative to demand can narrow even as production increases from the three major manufacturers swell the global net increase overall. Whether mass production proceeds as scheduled, and whether the fulfillment rate eventually moves out of negative territory, will be the next key points to watch in gauging DRAM's supply capacity.