Every few weeks, a new report lands claiming that most small businesses are now using AI. The numbers are always impressive. The methodology is usually buried.
This week, the U.S. Census Bureau published a different kind of data.
What the Government Actually Found
The Census Bureau's Business Trends and Outlook Survey (BTOS), a biweekly nationally representative survey, tracked AI adoption across U.S. businesses from December 2025 through early May 2026. The results are sobering.
Overall AI usage hovered between just 17% and 20% of all U.S. businesses during that period. Among very small firms -- those with four or fewer employees -- the number stayed below 20%.
The contrast with large firms is stark: 37% of companies with 250 or more employees reported using AI. For firms with 100 to 249 employees, the rate was 32%. For the smallest businesses, which make up the vast majority of the country's 33 million small business owners, the number barely moved over six months.
The Census data also noted something important: AI use increased among firms with at least 20 employees during this period, but "didn't change significantly" among those with fewer than 20. The smallest operators are not catching up.
The Survey vs. the Survey Problem
Compare that to the 2026 QuickBooks AI Impact Report, released earlier this month, which surveyed 34,000 small businesses and found that 77% reported using AI regularly. That's a number that's been cited widely across business press. The report also found that 41% saw revenue increases and 74% reported improved productivity.
The problem, as Forbes contributor Terdawn DeBoe laid out this week, is in how those gains were measured. More than 50% of the businesses surveyed described their improvement as a "general feeling." Less than half tracked specific metrics. Productivity numbers were self-reported, not based on time studies. Revenue gains were correlation, not controlled measurement.
In other words: businesses that pay for QuickBooks and opted into an Intuit survey believe AI is helping them. That's useful signal, but it's not the same as neutral, representative data.
The Definition Gap
Part of what's driving this divide is how "using AI" gets defined. The Census Bureau noted that it revised its own question last November -- from asking whether businesses used AI "in producing goods or services" to asking whether they used it "in any business function." That definitional change alone likely inflated apparent adoption rates across the industry.
When a survey counts using Grammarly, Google autocomplete, or an AI-assisted customer service chatbot as "using AI," adoption looks universal. When the question is about AI doing something material to business operations, the numbers drop sharply.
What This Means If You Run a Small Business
The Census data suggests most very small businesses are not yet running AI workflows in any meaningful operational sense. That's not a failure -- it's a realistic picture of where the technology actually is for most operators.
A few practical takeaways:
Don't benchmark against survey headlines. If a vendor report claims 77% of businesses are using AI and you're not, you're not behind. You may just be in the statistical majority of real-world operators.
The firms that are pulling ahead tend to have at least 20 employees. This likely reflects the capacity to assign someone to evaluate and implement tools, not just sign up for them.
Adoption isn't the goal -- measurable impact is. The QuickBooks report is worth reading precisely because it shows how hard it is to attribute outcomes to AI. Before you add another tool, decide how you'll know if it worked.
The hype cycle is real, and it benefits vendors to make adoption look universal. The Census Bureau doesn't have a product to sell. That's worth something.
Sources: U.S. Census Bureau, Business Trends and Outlook Survey, May 2026; 2026 QuickBooks AI Impact Report; Forbes, May 29, 2026