Small businesses are not backing away from AI. They are getting more disciplined about it.
That is the useful read from a Business Insider report published today showing that owners who adopted AI to save money are now treating it like any other operating expense, complete with guardrails, budgets, and a healthy amount of suspicion.
The basic story is familiar: firms under pressure are using AI to do more with fewer people. Sparkles Homes says automation helped it survive tariff pressure and keep a five-person team intact. Flint Avenue Marketing is using AI assistants to help answer phones and chase leads. That kind of workflow is exactly why small businesses have been among the fastest adopters of AI tools.
The new part is the bill.
Business Insider reports that a U.S. Chamber of Commerce survey found 58% of small businesses used AI in 2025, up from 23% in 2023. It also cites Ramp economist Ara Kharazian saying small firms that do adopt AI tend to use it more intensely than larger companies do. Ramp’s data, as reported by BI, shows small businesses spending a median of about $21 per employee on AI, compared with $11 across businesses of all sizes.
For the smallest firms, that starts to matter quickly. The same report says firms with 0 to 49 workers spent an average of $607 per worker in 2025 and expect that figure to reach $1,034 in 2026, based on an analysis from the Federal Reserve Bank of Atlanta. That does not mean AI is failing. It means the old pitch of "buy this tool and save immediately" is giving way to something closer to normal software management: usage caps, internal policy, and a spreadsheet that keeps getting larger.
That shift shows up in the anecdotes too. One owner told BI she accidentally spent $1,000 generating stock images before setting a daily spending limit. Another said she now keeps a monthly buffer in case prices rise. Those are not edge cases anymore. They are the signs of a tool graduating from experiment to infrastructure.
There is also a bigger strategic lesson here for Main Street: AI is not just a productivity hack. It is a dependency. Once it starts handling customer calls, drafting outreach, or powering basic admin work, the risk is not only that it gets something wrong. It is that a business becomes reliant on a system it does not fully control.
That is why the most practical response is not resistance. It is discipline. Use AI where it clearly saves labor, but track the cost per task, set hard limits, and make sure a human still checks anything customer-facing. The winners in this phase will not be the firms that use the most AI. They will be the ones that know exactly what it costs.
Sources: Business Insider, U.S. Chamber of Commerce, Federal Reserve Bank of Atlanta, JPMorganChase Institute