Seeing the news of soaring semiconductor stocks, not a few readers likely felt that the balance of power in the DRAM industry had already been rewritten. CXMT listed on the Shanghai Stock Exchange on July 27, 2026, with a valuation based on its offering price of roughly $85 billion, and its share price surged by about 470% on the first day of trading, pushing its market capitalization above that of the Industrial and Commercial Bank of China. But this capital-market victory doesn't necessarily translate into DRAM product competitiveness.
CXMT's production capacity has climbed to nearly 90% of Micron's, yet its market share by revenue is only about one-third of Micron's. Around the same time, SK hynix is moving forward with plans to supply next-generation LPDDR6 memory in the second half of the year. Which will move the procurement decisions of major Chinese smartphone makers—CXMT, which has won on capital, or SK hynix, whose actual capabilities haven't yet been converted from capacity into revenue share?
The Day It Overtook ICBC in Market Cap: Inside CXMT's $85 Billion Listing
On July 27, 2026, CXMT's share price rose about 470% from its offering price on its first day of trading on the STAR Market of the Shanghai Stock Exchange. The IPO is said to have raised approximately $8.6 billion to $9.8 billion for CXMT, bringing its valuation based on the offering price to roughly $85 billion (579 billion yuan). The range in the amount raised appears to stem from whether an over-allotment option was exercised, though even the source material doesn't settle on a single figure. After the share price surge, CXMT's market capitalization rose even further above its offering-price-based valuation, reaching approximately 3.3 trillion yuan (roughly $487 billion to $489 billion), briefly making it the largest company listed on mainland China by market cap, surpassing the Industrial and Commercial Bank of China.
Converting the offering-price-based valuation of $85 billion at the exchange rate as of July 28, 2026 (163.7 yen to the dollar) comes to roughly 13.9 trillion yen. It is itself unusual for a Chinese company specializing in semiconductor memory to list on the mainland market at this scale of valuation, and it reflects the market strongly pricing in the tailwind of AI-driven DRAM price surges.
But what the stock market priced in is future growth expectations, not current product competitiveness. As discussed below, CXMT's DRAM revenue share still trails far behind Samsung and SK hynix. Where the $8.6 billion to $9.8 billion raised gets deployed will be the dividing line for whether this gap can be closed. Whether it goes toward expanding production capacity or is concentrated on catching up in process development will greatly affect how quickly capital strength converts into technological strength.
The Key Held by 10.7Gbps LPDDR6, and SOCAMM2 as the Next Target
On March 10, 2026, SK hynix officially announced the completion of development for its 16Gb LPDDR6 DRAM using the "1c" process, corresponding to the sixth-generation 10nm class. The operating speed exceeds 10.7Gbps, offering a 33% improvement in data processing speed over the previous-generation LPDDR5X while cutting power consumption by more than 20%. The company has stated it plans to finish mass-production preparations within the first half of 2026 and begin supply in the second half.
Process generation naming is an internal benchmark indicating the degree of miniaturization; as the process advances from 1a to 1b to 1c, circuit density within the same area increases and performance per unit of power consumption improves. SK hynix's deployment of the latest 1c generation in LPDDR6 represents a renewal in power efficiency. A 20% reduction in power consumption directly affects a smartphone's battery life and heat management, with devices that run AI processing frequently benefiting the most. Because a higher clock speed can be sustained within the same power budget, this becomes a difference that can determine chip selection itself for device manufacturers.
In parallel, JEDEC is advancing the standardization of "SOCAMM2," a memory module specification for AI data centers based on LPDDR6. It targets a maximum capacity of 512GB, and SK hynix, Samsung, and Micron are all said to be pursuing this capacity target. A picture emerges in which SK hynix is simultaneously pursuing two use cases—power-efficient design for smartphones and large-capacity design for data centers—on the same LPDDR6 foundation.
What a Monthly Output of 350,000 Wafers Reveals: The Gap Between Production Capacity and Market Share
According to Omdia's tally, global DRAM market share by revenue as of Q1 2026 stood at 38.6% for Samsung, 28.8% for SK hynix, 22.4% for Micron, and 7.6% for CXMT. CXMT's share rose from 4.7% in the previous quarter, suggesting it had been rapidly increasing its presence even before its listing.
| Company | Projected Monthly Capacity (End of 2026) | DRAM Market Share (Q1 2026 Actual) |
|---|---|---|
| Samsung | 720,000 wafers | 38.6% |
| SK hynix | 595,000 wafers | 28.8% |
| Micron | 385,000 wafers | 22.4% |
| CXMT | 350,000 wafers | 7.6% |
It should be noted that production capacity figures are projections as of year-end, while market share figures are actual results from the January-March quarter, meaning the reference points differ. However, CXMT's production capacity has continued to expand rapidly over the past year, and even correcting for the difference in timing does not work in the direction of narrowing the gap between the two figures. Lining up these two numbers reveals a twist in CXMT's positioning.
In terms of production capacity, CXMT's monthly output of 350,000 wafers works out to 90.9% of Micron's 385,000 wafers. Yet in revenue share, against Micron's 22.4%, CXMT stands at only 7.6%, a ratio of just 33.9%. While the two companies are nearly on par in production capacity, CXMT's revenue share comes to only about one-third of Micron's.
Running the same calculation between SK hynix and Samsung yields a capacity ratio of 82.6% (595,000/720,000) and a market share ratio of 74.6% (28.8/38.6), a gap of only about 8 percentage points between the two. CXMT's gap between capacity ratio and revenue share ratio reaches 57 percentage points—a disparity that can't be explained away simply by the growing pains of an emerging company.
What this discrepancy likely indicates is that much of CXMT's production lines are still allocated to older-generation or lower-priced products. Even though CXMT is reported to have begun sample shipments of the latest LPDDR6, its mass production mainstay appears to remain general-purpose DDR4 and DDR5, and the low unit prices are directly reflected in its sluggish revenue share growth. Looking only at the capacity figures and concluding that the two companies are now on equal footing would overlook this gap in unit pricing.
Will Xiaomi Bypass Its Own CXMT? The Substance and Limits of a Single Report
Regarding where SK hynix's LPDDR6 will first be supplied, South Korean media outlet Herald Corp reported that the initial supply destination would be Xiaomi. In response, SK hynix stated in a report dated July 28, 2026, that "it is unable to confirm information related to specific customers," declining to officially confirm the supply destination.
The specialized tech outlet wccftech, which followed up on this report, classified the reliability of the information itself as "rumor" level, giving it a low rating of 1 out of 5 on an independent-corroboration-strength indicator and 55% on a plausibility indicator. This is an example of how, as information originating from a single South Korean media outlet spreads to international tech media, the thinness of the sourcing gets buried under a definitive-sounding headline.
If Xiaomi were indeed to adopt SK hynix's LPDDR6, it could be interpreted as a move to bypass its own domestic supplier CXMT for certain products. Behind this lies an incident reported in June 2026, in which CXMT had engineers from Huawei-affiliated semiconductor equipment company SiCarrier removed from its Hefei R&D facility. This is said to stem from friction with Huawei over DRAM pricing, and friction over price negotiations has also arisen within the domestic Chinese supply chain centered around CXMT.
SK hynix has not disclosed any information about the supply prices or preferential terms offered to Xiaomi. Nor has CXMT disclosed the actual yield rates or performance figures of its LPDDR6 samples. The information both companies keep silent on is exactly what would help determine whether CXMT, which leads on capital, or SK hynix, which leads on technology, ultimately gains the real advantage—but at this point, it remains a matter of speculation.
SK hynix is not the only one that stands to benefit from this dynamic. Other major Chinese smartphone makers such as OPPO and Vivo also have an incentive to diversify their supply chains rather than depend on a single vendor, at least until CXMT's technology generation catches up. The ones who stand to lose out are CXMT and Huawei, whose price-negotiating leverage is being weakened, and the key question going forward is how much this balance of power can be reshaped by the capital raised through the listing.
The 2018 Lesson of Fujian Jinhua: Can Capital Close a Technology Gap?
China's push for domestic DRAM production is not a first-time challenge. Fujian's state-owned enterprise Fujian Jinhua (JHICC) became subject to U.S. Department of Commerce export restrictions in 2018 over national security concerns, derailing its originally planned mass production of cutting-edge DRAM. Being cut off from dependence on U.S.-origin manufacturing equipment and technology dealt a major blow, and the company subsequently shifted course to producing commodity DDR4 to return to the market.
CXMT's current position differs from this precedent in its conditions. The roughly $8.6 billion to $9.8 billion raised through the Shanghai listing represents a depth of equity capital that Fujian Jinhua never had at the time. Production capacity has also climbed to nearly 90% of Micron's, making the risk of derailment from simple funding or equipment shortages relatively small.
But capital strength, while able to buy more equipment or factory floor space, cannot buy the time needed to close a full process generation. For SK hynix to reach the 1c-generation LPDDR6, it had to accumulate yield improvements through the 1a and 1b generations. For CXMT to shift its mass-production mainstay from DDR4 and DDR5 to LPDDR6 and DDR6, it will need to traverse that same learning curve on its own.
Full-scale commercialization of DDR6 is expected around 2028-2029, and CXMT is said to be moving ahead of the curve on supply chain transitions such as panel-level packaging. How much yield it can build up during this transition period will be the deciding factor in whether it can acquire technological capability commensurate with its $85 billion valuation. Japanese smartphone and PC makers also depend heavily on overseas manufacturers for DRAM procurement, making the competitive dynamic between CXMT and SK hynix an issue that could also ripple into domestic component procurement costs.