As AI use increases at work, many employees still choose not to use it: Gallup poll

Gallup’s February 2026 poll of 23,717 employed U.S. adults shows about 3 in 10 workers are frequent AI users, while roughly half use AI once a year or not at all. About 4 in 10 say their organization has adopted AI tools, and two-thirds of those users report a positive impact on productivity, but resistance remains high: 46% of non-users prefer working without AI and about 4 in 10 cite ethics, privacy, or skepticism about usefulness. The survey also shows rising displacement concerns, with 18% of workers saying their job is likely to be eliminated within five years, up from 15% in 2025.

Analysis

The more important takeaway is not adoption, but dispersion: AI is becoming a productivity enhancer for knowledge-heavy managers while remaining largely irrelevant or actively resisted in lower-complexity workflows. That creates a widening operating-margin gap inside the same industry between firms that can standardize AI-enabled workflows and those that cannot, with the fastest monetization likely in software, cybersecurity, and workflow automation rather than headline AI model providers.

A second-order effect is labor mix compression. If AI boosts output most in managerial, legal, health-adjacent, and tech roles, companies will quietly freeze backfills and reduce junior hiring before they announce large layoffs; that shows up first in slower wage growth, fewer entry-level openings, and better leverage on SG&A over the next 2-4 quarters. Conversely, service businesses may see limited ROI, which means AI capex there can destroy value if adoption is forced top-down without process redesign.

The market is probably underpricing compliance and governance spend. As workers and managers adopt consumer-grade tools in regulated settings, the bottleneck shifts from model quality to data controls, auditability, and identity management. That should support cybersecurity, data-loss prevention, and enterprise governance vendors even if pure-play AI enthusiasm cools; the winners are the picks-and-shovels around controlled deployment, not the broadest adopters.

The contrarian risk is that survey skepticism delays near-term productivity gains enough to create a sentiment air pocket in AI-adjacent software names over the next 1-2 quarters. But that weakness would likely be buyable if CIOs respond by mandating approved tools after a wave of high-profile errors or privacy incidents, which would convert optional experimentation into enterprise rollout.

Key Decisions for Investors

• Long PANW / CRWD on a 3-6 month horizon: regulated AI adoption should increase demand for governance, endpoint control, and data exfiltration defenses; look for a better entry on any AI software pullback.
• Pair long MSFT vs short lower-quality service-automation enablers: Microsoft benefits from distribution plus enterprise trust, while forced AI adoption is more likely to expose weaker workflow vendors with unclear ROI.
• Add to NOW or HUBS on weakness over the next 1-2 quarters: workflow platforms that embed AI into existing processes can capture productivity gains without requiring behavioral change; risk/reward is favorable if enterprise budget scrutiny intensifies.
• Avoid chasing pure AI model beta after a hype run; instead use any 5-10% selloff in enterprise governance and security names as a tactical entry, since adoption friction usually delays revenue by months but extends the spend cycle for years.
• Consider a relative-value short in labor-intensive service software / BPO exposure versus long AI-enabled enterprise software if adoption remains uneven; the trade works best where pricing power is low and management claims AI-led efficiency without measurable margin evidence.