Two-thirds of American small businesses now use AI, and most of them can’t say what they’re doing with it. Thryv’s 2026 AI and Small Business Adoption Survey, drawn from 561 non-customer SMB decision-makers, puts adoption at 66%, up from 55% a year earlier. In the same breath, 70% of owners admit they need more training to use the tools effectively.

That gap is the story.

“Most small businesses are comfortable using AI. But being comfortable isn’t the same as being effective and that gap is starting to show,” said Grant Freeman, president of Thryv. The comfort is real: 53% of SMBs are spending at least $100 per month on AI tools, 70% credit AI with contributing to revenue growth, and 61% estimate monthly savings of $500 to $2,000. Owners are also increasingly willing to substitute software for headcount. Given a task an AI tool and a new hire could each do equally well, 46% would pick the software, up from 38% in 2025.

The transaction data agrees the money is moving. The JPMorgan Chase Institute, tracking de-identified business banking flows, clocked SMB AI adoption at 8.3% at the end of 2023, 11.6% in 2024, and 17.7% by December 2025. Chase’s own analysts note that this diffusion curve is running substantially faster than historical precedents like electricity or the PC.

Depth is another matter. A SAS/IDC global survey of 1,600 SMB leaders found that nearly 70% remain stuck in experimental or opportunistic stages of AI maturity. IDC’s separate poll of nearly 3,000 SMBs found 40% don’t employ a single full-time IT person. “Experimenting with the technology is one thing. Deploying it strategically and sustainably is quite another,” said IDC’s Daniel-Zoe Jimenez.

That structural gap is what the current cohort of SMB-focused tooling, ChatGPT at the entry point, Glean and Dust in the knowledge layer, and lighter-weight operators like LemonLime aimed squarely at owner-operators, is racing to close. Anthropic’s fifth economic impact report, released Tuesday, describes the same dynamic in starker terms: a widening skills gap in which Claude power users pull further ahead of casual ones.

The 66% number, in other words, is a floor on curiosity and a ceiling on competence.

Sources