Chip stocks drop as AI leaders propose development slowdown

Semiconductor stocks tumble on AI pacing proposal

Global semiconductor and memory stocks declined Monday following a weekend essay by Anthropic CEO Dario Amodei calling for the AI industry to slow the pace of frontier model development. The proposal, which gained support from Sam Altman, Elon Musk, and Satya Nadella, comes after public resignations of safety staff at major AI labs concerned about existing safeguards.

Memory producers bore the brunt of the selloff. SK Hynix dropped 6.4% and Samsung Electronics fell 4.1% in South Korea, while Kioxia declined 6.4% in Japan. European chip stocks also opened lower, with ASML down 4.4%, Infineon and ASM International falling more than 5%, and STMicroelectronics shedding 3.5%.

Why it matters

Memory makers face greater exposure to AI spending shifts than equipment suppliers or IP providers. Unlike ASML, which carries a multi-year equipment backlog, or Arm, which operates with software-like margins, companies like SK Hynix, Samsung, and Micron have driven margin expansion through tight high-bandwidth memory supply and favorable pricing. Any pullback in AI capital expenditure could directly impact their earnings if weaker demand leads to rising inventories and softer pricing, though expanding inference workloads and existing supply agreements may provide near-term cushioning.

Analysts see limited spending impact

Bank of America analyst Vivek Arya characterized the developments as "noise" relative to secular AI market growth, projecting AI capital expenditure could surge threefold to over $3 trillion by decade's end. He cited robust demand signals including 100% network utilization, rising rental rates for older-generation chips, and intensifying competition between the U.S. and China, hyperscalers and neoclouds, and sovereign AI programs.

Bernstein analysts, while acknowledging the news could "further depress sentiment" on semiconductors, questioned whether a "pacing slowdown" necessarily implies reduced spending. They noted Amodei is not calling for a halt to training but rather a shift from "extremely fast" to "only somewhat fast" development—a pace that would still represent significant growth. The firm also pointed to rising AI demand from inference and agentic use cases, noting insufficient compute capacity exists even for current models.

Anthropic's own spending continues

Despite the pacing proposal, Anthropic's capital commitments show no signs of slowing. According to The Information, the company agreed to 14.8 gigawatts of computing capacity carrying approximately $517 billion in commitments over eleven months through August, with four deals signed since July. The Financial Times reported Anthropic expects a second consecutive quarter of adjusted operating profit, with inference gross margins exceeding 80% before distribution costs.

Bernstein maintains a positive outlook on AI infrastructure buildout, naming Nvidia, Broadcom, and semiconductor capital equipment makers as preferred sector plays.

These details were first reported by Investing.com.