Semiconductor Stocks Fall as AI Leaders Call for Slower Development Pace

Shares of semiconductor and memory chip manufacturers fell globally on Monday after a proposal from Anthropic CEO Dario Amodei to slow the pace of artificial intelligence development rattled investor sentiment. The call to "pace the frontier," which was reportedly echoed by other industry figures, follows concerns over AI safety and has sparked worries about a potential reduction in the sector's capital spending.

Global Sell-Off Hits Chipmakers

The reaction was swift across key Asian and European markets. In South Korea, memory giants SK Hynix and Samsung Electronics saw their shares fall by 6.4% and 4.1%, respectively. Japan's Kioxia also tumbled 6.4%.

European semiconductor firms opened significantly lower, according to market data. By 07:10 GMT, notable declines included:

• ASML: -4.4%
• Infineon: > -5%
• ASM International: > -5%
• BE Semiconductor: -4.8%
• STMicroelectronics: -3.5%

Market Impact and Analyst View

Ad

Analysts at Bernstein, led by Stacy Rasgon, noted the news has the "potential to further depress sentiment" for a sector that had already fallen roughly 19% from its June peak. The proposal raises concerns that a slowdown in training new, larger AI models could curb the voracious demand for high-performance chips.

Memory producers are seen as particularly exposed. The recent margin expansion for firms like SK Hynix, Samsung, and Micron has been heavily driven by tight supply and strong pricing for high-bandwidth memory (HBM), a critical component in AI accelerators. Any pullback in AI-related spending could directly impact their earnings, though existing multi-quarter supply agreements may provide an initial buffer.

A Slowdown, Not a Halt

However, Bernstein analysts cautioned that a "pacing slowdown" does not necessarily equate to a spending slowdown. They highlighted that Amodei's proposal suggests a shift from an "extremely fast" to a "somewhat fast" pace, which would still represent significant growth. The analysts also pointed to rising demand from inference and other AI applications, stating there is "already not nearly enough compute to satisfy demand for even the current models."

Adding a layer of complexity, Anthropic's own actions appear to contradict a spending freeze. According to The Information, the AI lab has made commitments for computing capacity valued at approximately $517 billion over the 11 months through August. Bernstein said it remains positive on the long-term AI buildout, naming Nvidia, Broadcom, and semiconductor equipment makers as its preferred investments in the space.