AI Fears Hit Software Again. Is the Repricing Overdue, or Overdone?

Enterprise software cannot catch a bid. On Tuesday, Morgan Stanley cut its ratings on Salesforce and Adobe and trimmed both price targets sharply, pointing to the worry that has trailed the group all year: that artificial intelligence will erode the subscription-software business faster than these companies can turn AI into new revenue. The stocks fell. So did much of the sector around them. ServiceNow reports results after today’s close, which puts the question back on the tape in real time.

The argument has a clean shape. One camp reads a structural break, in which AI agents and cheap generative tools chip away at seat-based pricing and the valuation premium software held for two decades. The other reads an overreaction, in which durable contracts and switching costs shield incumbents long enough to sell their own AI. Both camps can point to this week’s action. Neither has been settled.

Analysts keep marking software down

The bears have the price action and, lately, the analysts. Morgan Stanley moved Salesforce to equal-weight from overweight and pulled its target to $185 from $287. It cut Adobe to underweight and dropped that target to $240 from $365, citing slow progress turning AI features into booked revenue. The calls were not isolated. More Wall Street firms now rate Adobe and Salesforce a sell than at any point in years, and at least five had already trimmed Adobe since the start of June.

The tape backs the caution. Salesforce has fallen about 28% this year and ServiceNow near 30%, with Adobe down a similar amount, a slide that has erased a large share of the sector’s value since AI worries took hold. Median enterprise-software revenue multiples have compressed from more than six times sales at the end of 2024 toward the low single digits this spring, a repricing investors have taken to calling the SaaSpocalypse. The product threat is concrete. In Adobe’s core creative market, generative tools from Midjourney, Runway and OpenAI now do work that once required a subscription.

If software is doomed, why are chips soaring?

The bulls start with a contradiction. JPMorgan’s research team has called the selloff overblown and built on broken logic. Its reasoning: investors are betting at once that AI will gut software demand and that the companies selling AI are overspending to meet it, and both cannot hold. If new AI products were about to replace enterprise software wholesale, the chipmakers and cloud providers powering them should be winning. Instead many have sold off alongside the software names. Over the year the market has repriced the entire complex, including the names AI is supposed to reward.

Their second argument is the moat. Large vendors sit on proprietary, industry-specific data and multi-year contracts that are expensive and disruptive to unwind. Dan Ives of Wedbush has called the drop a generational buying chance, arguing AI will complement installed platforms well before it displaces them. And AI monetization has barely started. Salesforce is pushing consumption and outcome-based pricing through its Agentforce agents, a change that could lift revenue if adoption builds. Morgan Stanley’s own downgrade faulted the pace of that shift, not its direction.

What the data shows

Tuesday made the split visible in a single session. Technology led every sector, up 2.89%, yet the gain was far from evenly shared. Chipmakers drove it: Micron rose 12.17%, Intel 8.64%, AMD 8.11% and Arm 7.46%. The software names went the other way. CrowdStrike fell 3.70%, Adobe 3.23% and ServiceNow 2.52%, ranking among the day’s worst large-cap performers even as their own sector index climbed. That is the divide both camps describe, priced into one afternoon. Hardware feeds the AI build; application software carries the risk that the same wave passes it by.

What would settle it

The near term offers concrete tests. ServiceNow reports today, with IBM alongside it. Salesforce’s next quarterly update, and its Agentforce metrics, especially current remaining performance obligations, will show whether AI features are converting into bookings. Adobe’s coming results will do the same for its creative and document franchises against the new generative tools. The guidance and the net-retention figures will matter more than the headline beats. If AI-assisted products start expanding what customers spend, the moat argument gains ground; if seat counts and renewal rates slip, the structural case gets its evidence.

For now the market holds both views at once, punishing software while bidding up the hardware that feeds the AI it fears. Tuesday’s downgrades sharpened the bearish read. The switching costs and the early state of AI monetization anchor the other. The next few weeks of enterprise results, starting after today’s close, will show which reading the numbers support.

This is a general-market summary for information only — not investment advice, and not a recommendation regarding any security.