Firms with fewer than 20 employees barely moved on AI use between December 2025 and May 2026, while firms with 250 or more employees pushed past 37%, according to the Census Bureau's own Business Trends and Outlook Survey. That's not really a story about willingness. It's a story about who already has software built for AI to plug into, and who doesn't. If you're a non-technical founder running a lean, funded team, the Census Bureau's own numbers put you squarely in the group that isn't moving, and the fix isn't a chatbot subscription.
What the Census Bureau's own May 2026 numbers show
The Business Trends and Outlook Survey (BTOS) is a biweekly Census Bureau survey that asks businesses a direct pair of questions: did you use AI in any business function over the past two weeks, and do you expect to within the next six months. In a story published May 26, 2026, titled "Large Firms With at Least 20 Employees Biggest AI Users," the Bureau reported that overall AI use held between 17% and 20% across survey waves from December 2025 through the wave that closed May 3, 2026, with between 20% and 23% of businesses expecting to be using it within six months. Those are national averages, and averages are exactly what hide the number that matters for a small operation.
The gap sits right at 20 employees
The Bureau's own words on where the growth landed: AI use "increased among firms with at least 20 employees" over that stretch "but didn't change significantly among firms with fewer than 20 employees." Broken out by size as of the May 3, 2026 wave, 37% of firms with at least 250 employees reported using AI, 32% of firms with 100 to 249 employees said the same, and firms with four or fewer employees stayed under 20%. The line isn't gradual. It sits at a headcount most funded, non-technical founders are on the wrong side of by design, because a small team is the point of a lean seed-stage build.
Industry compounds it. In the same release, the Information sector reported 39.7% current AI use, with about 42% expecting to use it within six months. Retail Trade sat at roughly 14% current use and about 17% expected. A small business in a retail-adjacent category, e-commerce, direct-to-consumer goods, local services, is looking at the bottom of two gaps stacked on top of each other: size and sector, both pointing the same direction.
What this costs a funded, non-technical founder
None of this shows up as a single bad number on a dashboard. It shows up as a founder who's still the one reconciling returns by hand while a competitor two headcount bands up has that step running through a system nobody thinks about anymore. It shows up in a vendor questionnaire from an enterprise prospect that asks, as a checkbox item now, "how is AI used in your operations," with no honest answer beyond a general-purpose chat tool one person opens occasionally. The Census Bureau's own data says that gap between "we use AI" and "we don't have it near our actual operations" tracks headcount almost exactly, which means it isn't closing on its own while the company stays small.
Where this data doesn't mean what it looks like it means
Here's the part worth saying plainly, because it cuts against the easy reading of these numbers: most of the founders sitting in that under-20% figure aren't behind because they lack interest in AI. They're behind because most of them don't yet have software worth attaching it to. You can't put AI into an invoice-reconciliation workflow that still lives in a shared inbox and a spreadsheet, because there's no workflow there yet, just a set of manual steps a person carries in their head. Buying an AI feature before the underlying system exists is solving the wrong problem in the wrong order, and it's the single most common way a small, funded team burns its runway chasing a category that Census's own numbers say won't move the needle for a business its size anyway.
A worked example
Picture a 13-person direct-to-consumer coffee subscription company, about a year past its seed round, run by a founder with a marketing background and no engineering hire. Subscription pauses, address changes and refund requests all funnel through a shared support inbox, triaged by hand each morning. A prospective wholesale partner's vendor questionnaire asks how AI factors into order fulfillment and customer service. The honest answer is that it doesn't, not because nobody considered it, but because the underlying system, the thing an AI layer would actually read from and write back into, is a shared inbox, not software. Bolting a chatbot onto that inbox would answer the questionnaire's letter without touching what it's actually asking about: whether the operation itself is built to scale past the founder's own attention.
What actually closes the gap
The fix isn't an AI subscription. It's building the operational system first, subscription state, refund logic, fulfillment status, as real software with a database behind it, on a fixed scope sized to what the business actually needs this year. AI becomes worth adding only once that system exists to attach it to, the same lesson our look at Intuit's 2026 small business data found in the gap between businesses that merely use AI occasionally and the smaller share that built it into an actual recurring workflow. Skipping straight to the AI feature is how a small team spends its budget on the part of the Census Bureau's chart that isn't moving, instead of the part underneath it that would let it move.
If the honest read of your operation is closer to a shared inbox and a spreadsheet than a system, what it actually takes to scope a build like that is the more useful next read, and if you're already feeling that gap as a daily cost rather than a future risk, the signs it's become the actual bottleneck is worth a look before anything else. We scope this kind of build, fixed-scope, so an owner without a technical background always knows what they're buying, under custom software development, with transparent pricing and an itemized estimate within 48 hours of a call.
Sources
- U.S. Census Bureau: Large Firms With at Least 20 Employees Biggest AI Users (America Counts, published May 26, 2026)
- U.S. Census Bureau: Business Trends and Outlook Survey (BTOS) program page (accessed August 14, 2026)