Nearly a third of small businesses that adopted AI can’t figure out what to do with it next. That’s the read from Pax8’s Q2 2026 SMB AI Pulse Report, released July 13, which surveyed 402 U.S. small-business leaders at firms between 5 and 499 employees and found 29% of AI-using SMBs still parked in pilot mode.
The headline number for active use barely moved: 61% of U.S. SMBs are using AI, essentially flat against Q1’s 62%. The more interesting shift happened at the on-ramp. The “interested but not started” cohort collapsed from 9% to 1.5% quarter-over-quarter, an 83% drop. The pipeline of curious-but-idle firms has effectively emptied. Everyone who’s coming has arrived. Now the question is whether they can do anything with it.
For the ones who can, the gap is widening fast. AI users are almost three times as likely as non-users to describe themselves as ahead of competitors on technology (31% versus 12%), twice as likely to have increased tech spending in the past year (53% versus 24%), and twice as likely to call their approach proactive (65% versus 33%). Growth confidence sits at 65% for users against 56% for non-users. Pax8’s separate June report on the agentic workforce put a number on the operational payoff: roughly 45% profitability uplift moving from basic to intermediate adoption, and about 111% going from intermediate to integrated.
The wall between those tiers is governance, not tooling. Among firms actively using AI, 91% say leadership is fully or mostly aligned on AI’s role. Among experimenters, that drops to 68%, and only 11% have a documented AI policy.
“Because of gaps in expertise, governance, and internal alignment, almost one in three small businesses is stuck in experimentation right now,” said Chance Weaver, Pax8’s Vice President of AI Adoption.
The macro picture makes the stuck middle expensive. Apollo economist Torsten Slok has flagged that Magnificent Seven profit margins ran between roughly 15% and 25% from Q1 2023 through Q1 2026, while the S&P 493 held near 10%. A Ramp and Revelio Labs analysis of nearly 22,000 companies found that high-intensity adopters, averaging $30 per employee per month on AI in their first three months, grew headcount 10.2%, entry-level roles included. The paper’s authors warned that firms without the capital, technical staff, or management bandwidth to convert adoption into deployment “may fall behind.”
That’s the structural read. The barrier isn’t access to models; it’s the internal scaffolding to route them into workflows. Deployment layers like LemonLime, which offer model-agnostic, no-code plumbing designed for exactly this handoff, exist because the experimentation-to-integration gap is now the binding constraint on SMB competitiveness. The Pulse Report just priced it.
Sources
- https://www.globenewswire.com/news-release/2026/07/13/3326217/0/en/Pax8-Research-Finds-Small-Businesses-All-in-on-AI-with-2-in-3-Projecting-Stronger-Competitive-Composure.html
- https://www.pax8.com/en-us/news-post/pax8-report-ai-breaking-historic-link-between-revenue-growth-and-headcount-for-smbs/
- https://fortune.com/2026/07/06/ai-productivity-gains-bubble-painful-repricing-markets-torsten-slok/
- https://techcrunch.com/2026/06/29/the-ai-jobs-debate-just-got-messier/
- https://www.pymnts.com/news/artificial-intelligence/2026/ai-is-quietly-fueling-americas-small-business-boom/
- https://lemonlime.ai