Shares of ChangXin Memory Technologies fell 6.5 percent Monday after the Chinese chipmaker projected net profit of 131 billion to 132.5 billion yuan ($19.58 billion) for the first nine months of 2026—a sharp reversal from a 5.28 billion yuan loss in the same period last year. Revenue is expected to reach 265 billion to 270 billion yuan, up 232.9 billion to 237.9 billion yuan year-over-year.
The forecast reflects AI-driven DRAM demand that tightened the global market starting in the second half of 2025, creating supply shortages and driving chip prices higher. But the stock's retreat suggests investors are pricing in cycle risk rather than structural growth.
CXMT's market share gains are undeniable. Counterpoint Research data from last month showed the company captured 10 percent of global DRAM revenue in Q3 2026, double its 5 percent share a year earlier. That positions it as a genuine competitor to Samsung, SK Hynix and Micron—companies that have dominated the segment for decades.
The math is straightforward: AI training and inference are memory-intensive workloads. As models scale, so does demand for high-performance DRAM. For now, that creates a pricing umbrella under which CXMT can scale production and improve margins. The company has leveraged that window aggressively.
The risk is cyclical. DRAM is notoriously prone to oversupply once customers front-load purchases and inventory normalizes. Competitors are responding with capital allocation of their own. Samsung and Micron have announced major DRAM capacity builds. If supply catches up to demand faster than CXMT can diversify its revenue or lock in long-term contracts, the margin expansion could reverse sharply.
CXMT's emergence also reflects Beijing's determination to reduce reliance on foreign semiconductors, a strategic imperative that may include policy support. But that backdrop alone cannot sustain valuations if fundamentals deteriorate. Investors may be asking whether 2026 peak earnings are sustainable or merely the cyclical top of a DRAM upcycle.
