Why CXMT Shares Surged to a Record 61.80 Yuan

CXMT’s record run was driven less by a single confirmed change in earnings than by a powerful combination of AI-memory optimism, supply scarcity and China’s push for semiconductor self-reliance. On August 17, ChangXin Memory Technologies shares rose 12% to 61.80 yuan, taking the company’s market capitalisation to 4.13 trillion yuan and extending its position as China’s most valuable listed company.

The rally had already rewritten China’s stock-market rankings

CXMT’s August 17 move was the latest step in an extraordinary debut. On July 27, the company’s shares closed at 49 yuan after rising about 466% from the 8.66-yuan IPO price. That gave CXMT a market capitalisation of about 3.3 trillion yuan and immediately placed it at the top of China’s mainland stock market by value.

The stock then passed Tencent’s valuation on August 13, when CXMT’s market capitalisation reached about 3.54 trillion yuan. By the August 17 close, the share price was about 7.1 times the IPO price—an increase of roughly 614%—and the market value was approximately 590 billion yuan above the level at which it overtook Tencent.

Those calculations show why the August 17 gain mattered: it was not simply a continuation of a strong IPO. Investors were repeatedly raising the valuation they were willing to assign to China’s leading domestic DRAM supplier.

Why investors saw CXMT as an AI-memory winner

The central thesis was that artificial intelligence is creating a memory bottleneck alongside the better-known demand for computing power. AI data centres require large quantities of DRAM and storage, while manufacturers cannot add production capacity quickly enough to satisfy customers.

Micron said in June that AI spending was driving strong memory demand and that production constraints could last for at least two years. The company said customers had committed $22 billion to secure memory supplies. Further commentary in August indicated that data-centre customers could often obtain no more than half of the memory they wanted, even at very high prices.

For investors, that combination suggested a favourable memory-cycle setup: tight supply can support pricing, capacity utilisation and margins for established producers. The market projected that logic onto CXMT, whose strategic importance is amplified by China’s effort to build a more self-reliant semiconductor industry. CXMT has been described as central to that technology push and as China’s leading DRAM producer.

The resulting valuation was therefore an expectations trade. It reflected confidence in future demand, domestic strategic support and CXMT’s ability to expand its technology and production—not just the company’s near-term reported results.

SanDisk strengthened the broader memory-sector narrative

SanDisk’s August 13 investor day supplied another reason for investors to remain bullish on memory stocks. The company forecast mid-to-high-teens revenue growth for fiscal years 2028 through 2030, citing demand from the rapid build-out of AI infrastructure.

SanDisk also argued that AI inference and the growing use of KV cache are changing the memory hierarchy. Its view was that AI data centres will become substantially more storage-intensive, with the total available market for enterprise data-centre flash reaching 1.2 zettabytes by 2030.

That outlook helped shift the market’s focus from a narrow “compute chips only” story to a broader AI-infrastructure story in which memory and storage are strategic constraints. It also supported the idea that demand could remain strong for several years rather than being limited to a short-lived server upgrade cycle.

SanDisk said it had taped out its first die for High Bandwidth Flash and was working toward initial samples for customers developing AI inference devices. Its roadmap also highlights the importance of producing more bits from scarce wafer capacity, a factor that can reinforce the market’s focus on supply discipline.

Why the SanDisk read-through has limits

The comparison is useful, but it should not be overstated. SanDisk is primarily a NAND flash company, while CXMT is principally a DRAM maker. A positive outlook for enterprise SSDs and high-bandwidth flash does not directly translate into CXMT revenue or prove that CXMT will capture the same demand.

The connection was instead a sector-wide read-through. SanDisk’s forecasts suggested that AI infrastructure could support sustained demand across several types of memory. Micron’s comments suggested that supply would remain constrained, particularly as AI workloads increase demand for DRAM capacity and memory bandwidth. Investors then treated CXMT as a China-specific beneficiary of that broader cycle.

The most important caveat: valuation moved faster than fundamentals

CXMT’s surge showed how quickly a strategic semiconductor narrative can become embedded in a stock price. The company’s July debut, its overtaking of Tencent and its rise to 4.13 trillion yuan all occurred within weeks.

That does not establish that the market capitalisation is justified by current earnings. It shows that investors were discounting a much larger future opportunity: stronger AI-related memory demand, persistent supply tightness, greater domestic importance and successful technological execution.

In short, CXMT rose because investors began valuing it as a critical China-based AI-memory champion rather than as a conventional cyclical chipmaker. SanDisk and Micron helped validate the global memory thesis, but the direct financial impact on CXMT remained an expectation rather than a demonstrated result.