AI Pacing Debate Sends Chip Stocks Into Sharp Retreat

A warning about AI safety became a trade in semiconductors on Monday, with Intel falling as much as 7%, AMD losing roughly 6% and Nvidia retreating more than 3% before paring its decline.

The selling followed Anthropic CEO Dario Amodei’s weekend call to slow the pace of frontier-model development. OpenAI CEO Sam Altman endorsed the idea, while Elon Musk also voiced support. The message was not a call to abandon AI investment, but markets treated it as a challenge to the assumption that data-center spending will keep accelerating without interruption.

Amodei proposed giving independent evaluators ongoing access to frontier labs, coordinating safety standards among AI companies and eventually pursuing international agreements. He also warned that increasingly capable agent swarms could create severe cyber risks within months if safety work fails to keep pace.

That distinction did little to protect the companies supplying the infrastructure. The Philadelphia Semiconductor Index fell nearly 6% during the session, while memory names such as Micron and SanDisk also came under pressure. The concern is straightforward: if model releases arrive more slowly or hyperscalers pause new capacity, demand for accelerators, high-bandwidth memory and networking equipment may still grow, but at a less explosive rate.

“Pacing” is not the same as stopping. It does, however, put the spending cycle on a slower clock.

The market’s internal rotation underscored the point. According to Reuters, software stocks rallied as chip shares sold off, while Microsoft, Alphabet and Meta held up better than companies tied directly to the construction of AI infrastructure. The Nasdaq ultimately fell 0.6%, far less than its early decline, while the S&P 500 lost 0.5%.

That resilience suggests investors were not pricing an AI collapse. They were repricing the suppliers most exposed to uninterrupted capital expenditure.

Nvidia remains the central test. Its chips power much of the training and inference buildout, making the company the largest beneficiary of rising AI compute demand and one of the market’s most sensitive gauges of any change in spending expectations. AMD faces a similar read-through as it tries to expand its accelerator business, while Intel remains exposed to broader semiconductor-cycle concerns.

The next battleground will be corporate guidance. If hyperscalers continue lifting capital budgets, Monday’s retreat may look like a sentiment shock. If spending plans flatten, the pacing debate will become an earnings problem.