Memory Stocks Slump Pre-Market, Micron Drops Over 5% as Calls to Slow AI R&D Hit Chip Market

TradingKey - On September 14, storage stocks in the US market generally weakened in pre-market trading, with Micron Technology (MU) and SanDisk (SNDK) both falling more than 5%, SK Hynix (SKHY) dropping over 7%, and Western Digital (WDC) and Seagate Technology (STX) falling over 4%.

Source: TradingView

On the news front, executives from several leading AI companies have recently begun emphasizing that as artificial intelligence models become increasingly capable, related security risks and potential impacts are also expanding, making it potentially necessary for the industry to slow the pace of R&D and iteration for the most advanced AI models. This statement is becoming a new variable for market attention, prompting investors to reassess AI capital expenditures, model training demand, and the growth trajectory of the entire supply chain.

Among them, Anthropic CEO Dario Amodei published a lengthy article on Saturday, proposing that the company will further strengthen AI safety management and consider introducing more independent third-party evaluation mechanisms to conduct stricter reviews of model capabilities and potential risks. Meanwhile, he also called on the broader AI industry to adopt wider measures to appropriately control the pace of development for frontier models.

Amodei's statement quickly drew responses from other tech industry heavyweights. OpenAI CEO Sam Altman subsequently stated that OpenAI is willing to adopt one of the key recommendations, namely allowing independent evaluators to gain employee-like access to better understand and assess model safety risks. Elon Musk, who operates SpaceX (SPCX), also publicly responded, stating that Amodei is correct.

Over the past few years, AI companies have continuously increased computing power investments, accelerating the training and commercialization of new models. This has driven rapid growth in demand for data centers, GPUs, networking equipment, and high-performance storage.

In particular, products such as HBM, DRAM, and NAND have become important foundations for AI server expansion; consequently, storage company stock prices have benefited greatly from expectations of rising AI capital expenditures. However, if the pace of training frontier AI models undergoes an adjustment, the market will naturally begin to re-examine supply chain expectations previously built on continuous expansion.