Investment Announcements Spark Oversupply Fears
Nomura believes investor concerns over a potential slowdown in the artificial intelligence memory market have been exaggerated, arguing that recent investment plans by South Korean chipmakers and Meta’s (NASDAQ:META) cloud strategy do not signal weakening demand.
The brokerage noted that South Korean memory manufacturers and related companies recently announced long-term investment programmes worth approximately 4.8 quadrillion won ($3.5 trillion), including around 3.7 quadrillion won dedicated specifically to memory production.
The scale of these announcements has raised concerns among investors that the industry could eventually face excess supply.
AI Demand Continues to Outpace Supply
According to Nomura, the current market remains characterized by a significant shortage of memory products driven by rapid artificial intelligence adoption.
The analysts said semiconductor manufacturers continue to prioritize higher-margin high-bandwidth memory (HBM), leaving supplies of conventional DRAM and NAND products relatively tight.
Nomura also emphasized that the newly announced investments are unlikely to have a meaningful impact on supply for several years because large semiconductor manufacturing complexes require lengthy construction and development.
New Capacity Will Take Years to Arrive
The brokerage highlighted the Yongin Semiconductor Cluster as an example of the industry’s extended investment timeline.
Nomura estimates that the project, originally launched nine years ago, is unlikely to begin limited production before late 2027, illustrating that more than a decade can pass between initial investment decisions and commercially meaningful output.
Meta’s Strategy Seen as Positive for AI Demand
Nomura also rejected suggestions that Meta’s (NASDAQ:META) plans to commercialize excess computing capacity indicate softer demand for AI infrastructure.
Instead, the brokerage believes offering unused data-center resources is a logical way for major technology companies to improve returns on invested capital while expanding access for customers such as OpenAI and Anthropic.
The analysts added that lower computing costs could ultimately encourage broader adoption of artificial intelligence applications rather than reduce demand for AI hardware.