The Cheap Chinese Alternative Just Got Expensive

"China's largest DRAM maker, CXMT, is reportedly charging more than Samsung's roughly $1,240 price for comparable 64GB DDR5 server memory modules." — Reuters, via Wall Street Engine, July 25, 2026. CXMT has secured a five-year agreement worth more than $7 billion with ByteDance and another worth over $3 billion with Tencent. The company generated $7.5 billion in Q1 revenue, up 719% year over year, and is preparing for an $8.6 billion IPO on Shanghai's STAR Market.

The memory trade has been one of the most crowded positions in global markets. Bank of America's Fund Manager Survey showed 82% of managers long global semiconductors for three consecutive months, with zero respondents short. Samsung and SK Hynix together make up 55% of Korea's KOSPI index. The bull case was simple: AI needs memory, there isn't enough, and prices will keep rising. The bear case — the structural falsifier the framework has been tracking — was equally simple: CXMT would eventually produce enough conventional DRAM to break the shortage and crash pricing. China's state-backed champion, operating with subsidized capital and captive domestic demand, would undercut Samsung and SK Hynix the way Chinese solar panels undercut Western manufacturers. That falsifier just inverted. CXMT isn't undercutting the incumbents. It's charging more than Samsung for comparable products. Beijing steered state-owned buyers toward domestic suppliers, giving CXMT captive demand and pricing power simultaneously. The company that was supposed to be the commodity flood became another beneficiary of the shortage. The timing is significant. This pricing data emerged the same week China began mass-producing homegrown DUV lithography tools, giving its semiconductor industry equipment self-sufficiency at the conventional chip tier without needing ASML. And Apple is actively testing CXMT conventional DRAM for China-market devices. The supply chain for a self-sufficient Chinese memory ecosystem is forming, but the pricing behavior of that ecosystem is monopolistic, not deflationary. Meanwhile, the positioning that was built on the bull case has been obliterated — not by the fundamental falsifier, but by leverage. South Korea's KOSPI suffered its fourth worst session in history on July 28, falling 10% with two circuit breakers triggered. Samsung dropped 9%, SK Hynix 11%. Roughly 30% of Korean retail investors are in margin calls. The 2x and 3x leveraged ETF products that amplified the rally are being forcibly liquidated. Korea's finance ministry restricted leveraged ETF trading and imposed 24-hour market monitoring. The stocks have been cut in half. The fundamental case has arguably never been stronger. Micron's Q3 was a record quarter: $41.46 billion in revenue, 81% gross margins, filling only 50-66% of demand. Nvidia's CEO said last week that the company is "constrained in HBM memory, LPDDR memory, every part of the supply chain." SK Hynix's chairman said Jensen Huang asks for more chips every time they meet. IBM's CFO said the shortage won't alleviate soon. SanDisk disclosed $42 billion in minimum contracted revenue with price floors through 2030. The mechanism scorecard says memory pricing power is confirmed and strengthening. The price scorecard says the trade is in crisis. That gap — the widest the framework's dual scorecard has ever shown — exists because the damage is coming from leverage, not from fundamentals. Korean retail borrowed at 2x and 3x to ride a trade that was correct on the fundamentals and catastrophically wrong on the position sizing. The cheap Chinese alternative got expensive. The expensive Korean trade got cheap. And the most crowded position in global markets is being unwound not because the thesis broke, but because the leverage did.