A Google AI breakthrough is pressuring memory chip stocks from Samsung to Micron

Recent advancements from Google are shaking up the memory chip industry, raising concerns among investors regarding potential declines in chip demand. Shares of leading memory manufacturers, SK Hynix and Samsung, experienced significant drops of 6% and nearly 5%, respectively, in South Korea. Meanwhile, Japanese company Kioxia faced a nearly 6% decline, following a downward trend in U.S. firms Sandisk and Micron, which also saw declines in premarket trading on Thursday. On Tuesday, Google introduced TurboQuant, a novel compression technique that promises to slash the memory needed for operating large language models by a factor of six. This method focuses on optimizing the key value cache—an essential component that retains previous computations of an AI model, eliminating the need for redundant calculations. The aim is to boost the efficiency of AI models, a pivotal objective for research labs worldwide. This breakthrough has sparked fears among investors that the demand for AI memory chips, which are crucial for training extensive language models from companies like Google, OpenAI, and Anthropic, could dwindle. Matthew Prince, CEO of Cloudflare, referred to the research as "Google's DeepSeek," drawing parallels to the efficiency advancements made by the Chinese firm DeepSeek last year, which led to a notable sell-off in tech stocks. Despite the market's reaction, Ray Wang, a memory analyst at SemiAnalysis, argues that Google's findings may not necessarily decrease the need for chips. He emphasized that the value cache remains a critical bottleneck that, when addressed, could enhance both model and hardware performance. Wang noted that improving performance typically leads to increased memory usage, suggesting that more powerful models will require advanced hardware. While Thursday's stock declines raised eyebrows, a combination of sustained demand and supply shortages continues to favor the memory market in the long haul. Memory prices have skyrocketed, benefiting companies like Samsung, SK Hynix, and Micron, with Samsung's shares climbing nearly 200% over the past year and Micron and SK Hynix seeing increases of over 300%. Analysts suggest that the recent volatility in memory stocks is primarily due to profit-taking trends. Ben Barringer, head of technology research at Quilter Cheviot, pointed out that memory stocks have enjoyed a robust performance, and given the cyclical nature of the sector, investors are eager to capitalize on gains. He characterized Google's TurboQuant development as evolutionary rather than revolutionary, indicating that it doesn’t fundamentally change the long-term demand landscape for memory chips. In a market focused on risk reduction, even minor developments can prompt investors to reassess their positions.