Key points
- JPMorgan Chase Institute tracked actual payments for AI services across 4.6 million small business accounts from 2019 through 2025 and found 17.7 percent had become paying AI adopters by the end of 2025, up from 5.2 percent in 2023.
- The US Census Bureau’s Business Trends and Outlook Survey puts overall business AI use at 17 to 20 percent as of mid 2026, with the smallest firms, those with four or fewer employees, still under 20 percent and rising only slowly.
- Both of those measured figures sit far below the frequently quoted 89 percent “use AI in some capacity” survey statistic, which counts any use at all rather than a sustained, paid business practice.
- The real adoption gap is not about money. Employer firms adopted AI at 26.1 percent by the end of 2025 versus 15.3 percent for solo, nonemployer businesses, a pattern JPMorgan’s researchers tie to having extra hands to implement the tools, not extra cash to pay for them.
- Entry level AI spending fell roughly 60 percent between 2019 and 2024, from about $50 a month to about $20 a month, which is the main reason adoption has accelerated even as most individual small businesses still spend modestly.
The adoption number nobody actually agrees on
Search for small business AI adoption and the first statistic almost everyone repeats is that roughly 89 percent of small businesses now use AI in some capacity, a figure that traces back to a widely cited US Chamber of Commerce survey. Search a little further and a very different number appears. The Census Bureau’s Business Trends and Outlook Survey, a large, methodologically rigorous federal survey, put overall business AI usage at 17 to 20 percent between December 2025 and May 2026. For the very smallest firms, those with four or fewer employees, usage sat under 20 percent and barely moved during that stretch.
The gap is not a contradiction so much as a definitional trick. The 89 percent figure typically counts any use of any AI tool at all, including a free ChatGPT question asked once. The Census and JPMorgan Chase figures measure something narrower and more useful, sustained, often paid, business use. When the JPMorgan Chase Institute tracked de identified Chase Business Banking transactions across 4.6 million small business accounts from 2019 through 2025, looking for actual payments to AI services rather than survey answers, it found 17.7 percent of small businesses had become AI adopters by the end of 2025, up sharply from 5.2 percent in 2023 but nowhere near the headline 89 percent.
Why entry costs collapsing matters more than any single tool
The more interesting story in the JPMorgan Chase data is not the adoption percentage itself, it is how fast newer businesses are reaching meaningful adoption compared with older cohorts. A business that started in 2019 took 77 months, more than six years, to reach 10 percent AI adoption within its cohort. A business that started in 2025 reached that same 10 percent mark in about six months, roughly thirteen times faster. The researchers point to falling entry costs as the main driver. Businesses that adopted AI in 2019 or 2021 typically started paying around $50 a month. Businesses adopting in 2024 started around $20 a month, a 60 percent drop in entry price over five years, largely due to inexpensive generative AI subscriptions that did not exist in the earlier period.
That price collapse also changed how many tools a typical adopter uses. In 2019, 89 percent of AI paying small businesses paid for exactly one AI service. By 2025, that share had fallen to 72 percent, while the share paying for two services rose to 18 percent and the share paying for three or more reached 9 percent. Small businesses are not just adopting AI faster, they are increasingly stacking multiple tools rather than betting everything on one.
The adoption gap that is not about money
Perhaps the most useful finding for an actual small business owner deciding whether to invest time in AI tools is what separates adopters from non adopters within revenue bands. JPMorgan Chase’s data shows employer firms, meaning businesses with at least one paid employee besides the owner, adopting AI at 26.1 percent by the end of 2025, compared with just 15.3 percent for nonemployer, solo operator businesses. That gap holds even when comparing firms at the same revenue level. Among the lowest revenue businesses, those earning under $250,000 a year, employer firms still adopted at 27.6 percent against 14.3 percent for solo operators in that same revenue band, and small employer firms actually out adopted large nonemployer firms, 27.6 percent versus 19 percent.
The researchers’ read on this is worth sitting with. It is not that solo founders lack the money for a $20 a month subscription. It is that a solo operator running every function of a business alone often lacks the spare hours to evaluate tools, build a workflow around one, and stick with it long enough to see a return, while a business with even one extra employee has more capacity to absorb that setup cost. A related 2025 NFIB survey found only 48 percent of nonemployer businesses reported meaningful familiarity with AI at all, against 55 percent for businesses with one to nine employees and 82 percent for businesses with fifty or more.
“Small employer firms adopted at higher rates than large nonemployer firms, 27.6 percent versus 19 percent, suggesting that having access to additional human capital to implement and integrate AI may be more important than financial resources alone.” JPMorgan Chase Institute, Understanding AI Use by Small Businesses, 2026
What time savings actually means in the surveys
Separate from the adoption data, industry surveys on time savings converge on a consistent range even though methodologies differ. Multiple 2026 small business surveys report that AI using business owners and managers save more than seven hours a week, while the average AI using employee across a small business saves closer to 5.6 hours weekly. About 58 percent of small business AI users report saving more than 20 hours a month in total.
These figures are self reported, meaning a business owner is estimating time saved rather than a researcher measuring it directly, which is a real limitation covered further down. But the figures are directionally consistent enough, and repeated across enough independent surveys, that the general order of magnitude, several hours a week for a typical regular user, is a reasonable planning assumption. The more useful exercise for an individual business is not adopting someone else’s average, it is applying the same logic to a specific task. If a task took 45 minutes twice a day and an AI tool cuts that to 10 minutes, that is roughly 70 minutes saved daily, or just under six hours a week, a number a business owner can verify for themselves within two weeks of tracking.
What revenue impact actually means in the surveys
The revenue side of this story needs the same scrutiny as the adoption side. Surveys report that 91 percent of small businesses using AI say it boosts their revenue, and about two thirds report the tools save them between $500 and $2,000 a month, with an average reported return on investment of 3.7 times the amount spent on AI tools. Businesses that have adopted AI are also reported to be nearly twice as likely to show year over year growth compared with non adopters.
Every one of those figures is worth reading with the same caution applied to the adoption headline earlier. A business owner reporting that AI “boosts revenue” is describing a perception, not an audited change in a profit and loss statement, and a business that was already growing and sophisticated enough to adopt AI early is a different kind of business than one that was not, which makes the growth correlation partly a reflection of who adopts rather than proof that adoption alone caused the growth. None of that means the underlying effect is fake. It means the honest version of this claim is that AI using small businesses report better outcomes, correlated with adoption, not that AI has been proven through controlled measurement to cause a specific revenue increase for a typical business.
| Source | What it measures | Headline figure | How it was measured |
|---|---|---|---|
| US Chamber of Commerce | Any AI use, ever | 89% | Survey, self reported, broad definition |
| US Census Bureau BTOS | Current business AI use | 17 to 20% | Large scale federal survey, narrower definition |
| JPMorgan Chase Institute | Paid AI service adoption | 17.7% | Actual bank transaction data, 4.6 million firms |
| Small business ROI surveys | Perceived revenue and time impact | 91% report revenue gain, 3.7x ROI | Survey, self reported, no audited financials |
A framework for measuring your own numbers
Rather than adopting an industry average as a target, the more reliable approach for an individual small business is a short measurement exercise. First, pick the single task that currently takes the most recurring time, the same starting point our tool focused guide to AI agents for small business recommends before choosing any specific product. Second, time that task manually for a normal week to get a real baseline rather than a guess. Third, run the AI assisted version of that task for two to four weeks and track the actual time spent, including corrections and review, not just the AI’s first draft time. Fourth, multiply the hours saved by an honest hourly value of the owner’s or employee’s time to get a dollar figure, and compare that against the tool’s monthly cost. That comparison, run on one real workflow, is worth more than any industry wide statistic for deciding whether a specific tool earns its place in a specific business.
Honest limitations in this data
Every figure in this piece has a real limitation worth stating plainly. The JPMorgan Chase transaction data is the most rigorous source here because it measures actual payments rather than self reported behavior, but it only captures paid AI services and explicitly does not count free tool usage or AI features embedded inside other software, meaning true usage is almost certainly higher than 17.7 percent even by JPMorgan’s own account. The Census Bureau’s BTOS figures are survey based and can shift from one release to the next as methodology and response rates change. The revenue and time savings statistics come from industry surveys with self selected respondents, meaning businesses satisfied enough with AI to keep using it are more likely to respond positively than businesses that tried it and quietly dropped it. And the correlation between AI adoption and business growth cannot be cleanly separated from the fact that more capable, better resourced businesses are also more likely to be early adopters of any useful new tool, AI included.
What this means in practice
None of this argues against small businesses using AI. The JPMorgan Chase data shows real, measured, accelerating adoption, real spending growth among committed users, and a genuine collapse in the cost of getting started. What it argues against is treating a headline adoption or ROI statistic as a reason to feel behind, or as proof that a specific tool will produce a specific result for a specific business. The businesses actually pulling ahead, based on this data, are not the ones chasing the highest reported ROI number. They are the ones with enough hands on deck to properly configure a tool, who measure their own time and cost honestly, and who treat the setup investment as real work rather than something a free trial handles automatically.
For a deeper look at which specific tools handle which small business functions, and realistic pricing for each, our companion guide on the best AI agents for small business covers customer support, sales, scheduling, content and finance tools in detail. This piece exists to answer a different question first, whether the numbers behind the small business AI story hold up, so that the tool decision that follows is grounded in an honest baseline rather than an inflated one.
Frequently asked questions
What percentage of small businesses actually use AI?
It depends on the definition. Surveys asking about any AI use at all report figures as high as 89 percent. More rigorous measurement, including the US Census Bureau’s Business Trends and Outlook Survey and JPMorgan Chase Institute’s transaction based research, puts sustained or paid business use closer to 17 to 20 percent as of 2026.
How much time do small businesses actually save using AI?
Self reported industry surveys put average time savings around 5.6 hours a week for a typical employee and over 7 hours a week for owners and managers, with 58 percent of regular users reporting more than 20 hours saved monthly. These are self reported estimates, not independently measured figures, so treat them as a rough planning range rather than a guarantee.
Does AI actually increase small business revenue?
Surveys report that 91 percent of small business AI users believe it boosts revenue, and adopters are nearly twice as likely to report year over year growth. Both figures describe perception and correlation rather than an audited, causal measurement, since businesses that adopt AI early also tend to be more resourced and growth oriented to begin with.
Why do solo business owners adopt AI less than businesses with employees?
JPMorgan Chase Institute’s data shows employer firms adopting AI at roughly twice the rate of solo, nonemployer businesses, even within the same revenue band. Researchers attribute this to available time and human capital for implementation, not financial resources, since even small businesses spending under $250,000 a year showed the same employer advantage.
How much does it cost to start using AI as a small business?
Entry level AI spending has fallen sharply, from a median around $50 a month for businesses that adopted in 2019 or 2021 to around $20 a month for businesses adopting in 2024, largely due to low cost generative AI subscriptions that did not exist in the earlier period.
See the underlying research
Read JPMorgan Chase Institute’s full transaction based study, and see our tool by tool guide for choosing the right AI agent for your business.