Micron, Nvidia and other chip stocks fall after tech leaders call for an AI slowdown. Here's what to know.

By Steve Goldstein and Britney Nguyen

Analysts don't expect the push for more responsible AI development to derail data-center spending

Micron's stock is falling Monday along with the broader chip sector.

The calls from Anthropic CEO Dario Amodei and rival executives to slow artificial-intelligence development hurt chip stocks on Monday, as investors assessed whether a more measured approach to AI would impact spending on hardware.

The iShares Semiconductor exchange-traded fund SOXX fell 5.6% on Monday. Shares of Micron Technology (MU) slid 5.3%, shares of Marvell Technology (MRVL) dropped 7.3% and shares of Nvidia (NVDA) retreated 3.4%.

Neocloud stocks got hit hard as well, with Nebius Group (NBIS) down 5.5% and CoreWeave (CRWV) off 6.8%.

Need to Know: AI has been carrying the stock market. An industry pause could pull the rug out, warns this Wall Street giant.

Whether the sharp moves lower are justified is a matter of debate on Wall Street.

"This is not a halt: the benefits for growth, healthcare and democracy still outweigh the tail risks; AI adoption and token trends remain intact; excessive slowing risks ceding leadership in AI; and the administration remains pro-AI," wrote Federico Manicardi and Victoria Campos of J.P. Morgan's international trading desk.

Jefferies equities trading analyst Jeffrey Favuzza noted that the most recent worries over AI safety are coming at a time when the chip sector is already meaningfully off its highs.

He also said there isn't "a clear sign that spending would be moderating" from the recent AI safety commentary, and noted that Anthropic's Amodei "explicitly called out this doesn't mean they are halting model training."

If AI training were to slow down, said Bernstein analyst Madison Rezaei, data centers in rural areas would see a decline in demand, given that many of those facilities "have been purpose-built" for training, which does not depend on the near-instant movement of data to work.

But as inference, or the process by which an AI model makes predictions based on what it has learned, commands an increasing share of data-center workloads over time, that creates additional debates. Rezaei noted that investors will be questioning how fast demand shifts from training to inference.

In her view, "the answer is not straightforward," as the profitable and widely adoptable use cases are not yet known by the market.

The rise of agentic AI, or systems that can run tasks with little to no human prompting, "are likely to need lower-latency infrastructure," Rezaei said, meaning those workloads will need data to move instantaneously.

Meanwhile, shares of software companies rose Monday, with cybersecurity stocks particularly strong.

CrowdStrike's stock (CRWD) was up nearly 14%, while Palo Alto Networks shares (PANW) jumped 13%.

-Steve Goldstein -Britney Nguyen

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.

(END) Dow Jones Newswires

09-14-26 1656ET