Investing.com - Enterprise software names are pointing sharply higher in pre-market trading on Monday as semiconductor stocks face broad selling pressure amid calls for broader slowdown in the AI arms race.

Anthropic CEO Dario Amodei added to concerns around the pace of AI development over the weekend, publishing an essay that proposed “pacing the frontier” and slowing improvements in advanced AI capabilities. The comments followed a series of public departures by safety and alignment staff at major foundation-model labs, amid broader debate over whether existing safeguards are sufficient.

Amodei’s call was echoed by industry figures including Sam Altman, Elon Musk and Satya Nadella, potentially adding further pressure to sentiment around AI-linked stocks.

Recent AI winners like Marvell and Intel are down 5.5% and 4.9%, respectively. On the other hand, CrowdStrike is leading the SaaS group with shares up 4.5% in pre-market Monday trade. The divergence is shaping up as one of the clearest expressions yet of a pair trade that hedge funds have discussed for several quarters: long enterprise SaaS, short AI hardware suppliers.

The thesis rests on a view that hyperscaler spending on AI infrastructure is plateauing while enterprise adoption of AI-powered software features continues to expand, insulating SaaS vendors from the hardware cycle while punishing chip makers and equipment suppliers.

ServiceNow and Atlassian are each indicated up 3.5% ahead of the open, with Workday and Adobe up 2.8%, and Intuit rising 2.3%. These names are the most direct equity expression of the rotation narrative: they sell recurring software subscriptions to enterprises embedding AI into workflows, a revenue stream that is largely decoupled from whether Nvidia ships another generation of training clusters.

On the other side of the trade, the damage in semiconductors and equipment is broad. Applied Materials fell 4.6%, Micron is seen down 4.4%, while major chipmakers AMD, Broadcom, and Nvidia are down 4.1%, 3%, and 2%, respectively.

The AI trade had for most of the past two years lifted both software and semiconductor stocks simultaneously, as investors priced a single unified buildout of AI capacity. That correlation has begun to fracture as concerns mount over possible overcapacity in data centers and signs of slower model-training investment from the large cloud providers.

Software vendors are increasingly viewed as downstream beneficiaries of AI adoption, exposed to enterprise budget cycles rather than hyperscaler capex decisions.

Third-quarter earnings season for large-cap technology companies is approaching, and any guidance commentary from Microsoft, Alphabet, Amazon, or Meta Platforms on AI infrastructure spending plans will either validate or unwind the slowdown thesis.

A hawkish signal from the Federal Reserve at its upcoming FOMC meeting could also complicate the SaaS rally: high-multiple software stocks carry significant duration risk, and any shift in the rate outlook would pressure valuations precisely as the group is attempting a breakout from semiconductor correlation.