CXMT Eyes $9.8B STAR Market IPO to Fund AI-Focused Memory Expansion

China's leading domestic memory-chip maker, ChangXin Memory Technologies (CXMT), has sharply raised the size of its planned Shanghai STAR Market listing, now aiming to raise 66.6 billion yuan ($9.8 billion) — nearly double its earlier target of 29.5 billion yuan. The IPO is priced at 8.66 yuan per share, with listing expected on July 27, 2026. Proceeds include about 13 billion yuan for DRAM technology upgrades, 7.5 billion yuan for wafer production-line improvements, and 9 billion yuan for R&D. CXMT plans to ramp high-bandwidth memory (HBM) wafer output from roughly 5,000 wafers per month in 2025 to 30,000–55,000 wafers per month by 2026–2027. By the end of 2026, it is targeting total monthly wafer capacity of 350,000 units across its operations. HBM back-end packaging production is scheduled to start in late 2026. Founded in 2016 out of Hefei's state-backed "506 Project" — launched after China's failed attempts to acquire overseas DRAM makers — CXMT remains closely tied to government-linked capital. Qinghui Jidian, an entity associated with Hefei, holds about 21.67% of the company. After years of support and extended losses, CXMT turned profitable in 2025. As of Q2 2025, CXMT held around 4% of the global DRAM market, ranking as the world's fourth-largest DRAM producer. For crypto and digital-asset investors, a major memory capacity buildout could alter the supply constraints that underpin AI compute scarcity narratives. Projects such as Render, Akash, and io.net rely in part on tight availability of GPUs and high-performance computing resources; additional HBM supply could gradually ease AI infrastructure bottlenecks. The expansion also unfolds under U.S. export controls aimed at restricting China's access to advanced chipmaking tools. For miners and AI compute providers operating across jurisdictions, a more fragmented chip supply chain introduces both risks and opportunities. Market participants will be watching whether CXMT can narrow the technology gap with incumbent leaders: success could reset global memory pricing, while delays would extend the runway for AI compute scarcity theses.