"No Reasons to Own"? Software Stocks Drop as New AI Tool Shakes Confidence
Software stocks started 2026 on the back foot. A fresh AI service from Anthropic-released Jan. 12-reopened the debate on whether traditional software models can hold their ground. Intuit fell 16% last week, with Adobe and Salesforce down more than 11%.
A Morgan Stanley software-as-a-service basket is off about 15% year to date, after an 11% slide in 2025. That's the weakest start since 2022, based on data compiled by Bloomberg. Sentiment is defensive, and buyers are scarce.
What spooked investors
Anthropic's Claude Cowork, a "research preview," can turn screenshots into spreadsheets and draft reports from scattered notes. Built quickly-largely with AI-it's the kind of capability that pressures high-price, task-based software. As one tech specialist put it, many buyers see no catalysts for a near-term re-rate.
"The pace of change is about as fast as I can remember," said Bryan Wong of Osterweis Capital Management. Translation: forecasting growth is harder, risk premiums rise, and multiples compress.
Anthropic has momentum, and that keeps the threat top of mind for generalist portfolios that already reduced software exposure in 2025.
Narrative vs. execution
Recurring revenue and high margins used to command premium valuations. Today, fears of AI agents automating core tasks overshadow those strengths. ServiceNow and peers trade at multi-year lows even as the Nasdaq 100 hovers near records.
Most incumbents haven't shown meaningful AI traction in reported numbers. Salesforce highlighted Agentforce adoption, but it hasn't moved revenue materially. Adobe added generative features across creative tools yet avoided updating key AI metrics in its December quarter.
Growth math is shifting
Bloomberg Intelligence projects software and services earnings growth in the S&P 500 to slow to ~14% in 2026, from ~19% in 2025. Meanwhile, semiconductors look stronger. Nvidia and peers have clearer revenue visibility tied to hyperscaler AI spend from Microsoft, Amazon, Alphabet, and Meta.
BI estimates sector profit growth of ~45% in 2025, accelerating to ~59% in 2026 for semiconductor-linked names. "There's more certainty about their growth given their customers," said Jonathan Cofsky of Janus Henderson. That clarity is winning flows.
Valuations: cheap for a reason-or an entry point?
The Morgan Stanley SaaS basket trades at ~18x next-twelve-month earnings, the cheapest on record and far below the ~55x average over the past decade. The old logic-predictable subscriptions deserve premium multiples-is under review.
As Wong noted, it's tough to price a business competing with 24/7 AI agents that can ship deliverables in a day. The market wants proof that incumbents can monetize AI, not just ship features.
What could change the story
- Revenue lift from AI add-ons: Clear pricing, attach rates, and upsell that show up in ARR and net retention.
- Productivity ROI for customers: Documented seat expansion tied to measurable cost savings or time-to-value.
- Distribution and data moats: Evidence that proprietary datasets or embedded workflows slow agent displacement.
- Shorter sales cycles: If AI use cases become standardized SKUs, cycles and procurement friction should improve.
- Capital allocation: Smart M&A, disciplined buybacks at depressed multiples, and targeted AI infra spend.
Why chips still lead-for now
Visibility beats hope. Chipmakers have orders linked to AI infrastructure, while software has promise without consistent monetization. That gap in certainty explains performance dispersion.
For benchmarking, keep an eye on the Nasdaq 100 versus high-quality software names. If software lags into earnings while estimates stabilize, setup improves.
Practical playbook for finance teams
- Track disclosures: ARR from AI products, ARPU uplift, AI seat mix, and net revenue retention by cohort.
- Watch gross margin: Agent costs, inference spend, and data licensing can pressure unit economics.
- Stress test TAM: Identify modules most exposed to agent automation (e.g., routine reporting, reconciliations).
- Prioritize "system of record" vendors: Stickier data gravity and workflow depth tend to resist displacement.
- Listen for customer evidence: Case studies with quantified savings, not anecdotes.
Bull and bear cases in one glance
- Upside: Faster-than-expected AI monetization, improving sales cycles, and stabilizing estimates spark multiple expansion from trough levels.
- Downside: Agents commoditize mid-tier workflows, growth decelerates into 2H26, and valuation support weakens further.
- Timing signals: Guidance that breaks out AI revenue, rising attach rates, and cohort retention recovery.
Where the street stands
Some firms see opportunity forming. Barclays expects software to "finally catch a break" in 2026 as spend holds and valuations stay attractive. Goldman Sachs argues that AI adoption could expand total addressable markets rather than replace them. D.A. Davidson thinks narrative has outrun fundamentals, making selective re-entry compelling.
As Chris Maxey of Wealthspire put it, existential AI worries won't vanish soon. But we're getting closer to a setup worth doing the work on.
If you're upskilling your team
Finance leaders leaning into AI workflows can accelerate experimentation and ROI tracking. For curated, practical resources, see AI tools mapped to finance use cases here: AI tools for Finance.
Your membership also unlocks:
