Picture the December board meeting at a $120 million manufacturer. A director who also sits on another company’s board says, “I keep reading that most of our peers already use AI. Are we behind?” The CEO has about ten seconds to answer, and either bad answer costs money. A shrug looks complacent. A panicked yes buys a year of expensive pilots that nobody owns.
The question now has a government number behind it. On November 5, 2026, the U.S. Census Bureau published the latest results of its Business Trends and Outlook Survey, a biweekly poll of American businesses. For the two weeks ending October 18, 40.0 percent of companies with 100 to 249 employees, and 44.1 percent of those with 250 or more, said they had used AI in at least one business function. Five survey cycles earlier, in the two weeks ending August 9, the figure for the 100 to 249 group was 30.3 percent.
We think the number is useful and the question it invites is the wrong one. This article covers what the number measures, what it leaves out, and what a CEO can tell a board that is better than yes or no.
What Does the Number Say About Companies Your Size?
Census sorts companies by headcount, not revenue, so it is only a rough match for a mid-market company with $20 million to $500 million in sales. Within that limit, the pattern is plain. For the two weeks ending October 18, AI use rose with company size: 22.8 percent of companies with one to four employees, 29.2 percent of those with 20 to 49, 33.2 percent of those with 50 to 99, 40.0 percent of those with 100 to 249, and 44.1 percent of those with 250 or more. The survey’s own margin of error on the two largest groups is about four points and three points.
The survey also asked what companies expect. Among those with 100 to 249 employees, 47.2 percent said they expect to use AI in some business function over the next six months, and among those with 250 or more, 53.6 percent did. If those plans hold, about half of midsize companies will be using AI by spring, and “we have not started” will be a position that fewer of your peers share.
One note on wording. Census’s current questionnaire calls the technology “Super Intelligence,” with AI in parentheses as its former name. We use AI here because that is the word your board uses.
How Fast Is It Moving, and What Is the Number Hiding?
In the Census Bureau’s own write-up from May 2026, 32 percent of companies with 100 to 249 employees reported using AI in the period ending May 3. The October reading of 40.0 percent is a climb of about eight points in under six months. One reading can swing, so treat the direction as firm and the exact size of the jump as softer.
Now the part the headline leaves out. The question asks whether the business used AI in any of its business functions in the last two weeks. Census widened that wording in November 2025 to include light uses such as drafting emails, so a yes can mean one employee with one chatbot. A review of the same data by the Federal Reserve Bank of Minneapolis, published May 29, 2026, described use as largely experimental and limited to specific tasks, and mostly adding to work employees already do rather than replacing it. The same review found wide differences by industry in early 2026, from nearly four in ten firms in information and professional services to about 12 percent in manufacturing. Your industry’s number matters more than the national one.
So a company can be in the 40 percent and have nothing to show for it. Picture a distributor that bought a few licenses, told nobody what to use them for, and has no owner and no starting measurement. It counts as an AI user in the survey. Now picture a peer in the other 60 percent with one well-run project that cut quote turnaround from three days to one. The survey counts that company as a non-user, and its customers may notice the difference before its board does. The number measures a tool, not a result.
Is “Are We Behind?” the Right Question?
Our view is that it is not, for a reason that has little to do with technology. A board asking whether you are behind is really asking three things: are we exposed, are we spending sensibly, and does management know what is going on? An adoption percentage answers none of them. It also invites the wrong response, which is to chase the number by buying tools. As we argued in why efficiency from a widely available tool will not be your edge, the tool is not scarce. What your company builds with it is.
There is a second reason the question misleads. In our experience, employees at companies with no AI program often use public AI tools anyway, on their own, with company information. If that is true at your company, you are already in the 40 percent, with no owner, no rules and no measurement, which is the worst version of being there. Finding out is a management task, and it does not require a technologist.
What Should You Tell the Board Instead?
We would give a board three things on one page, each of which someone other than the CEO can answer. This is our judgment, not a standard.
First, where AI is used in the company today, officially and unofficially, and by whom. A list of departments and the tools they touch is enough. Second, one business measure that has moved because of it, with the starting number and the date it was taken. Hours to close the month, days to answer a quote, cost per support ticket and dollars of new orders per sales rep all qualify. If no measure has moved, say so. That is the honest answer to “are we behind,” and it is more useful to a board than a percentage.
Third, what has been decided about risk. That covers what company data may go into public tools, what your insurer and your largest customers have asked about it, and which laws apply to you. State laws on AI in hiring start January 1, so ask your counsel which of them reach you. A board that hears this much is less likely to keep asking about the percentage, because it has something better.
In our view, boards are moving the same way. They want proof more than activity reports, which we covered in what your board actually wants to see on AI. If your directors cannot judge the answers themselves, a strategist in an advisory chair can help without adding a director.
What Does It Cost, and Who Should Own It?
Ask your CFO for the total you spent on AI in the last twelve months: licenses, usage fees, outside help, and the staff time of the people who run it. Many companies we meet cannot produce that figure in a day. That gap is the first thing to fix, because a board cannot judge a return on a cost nobody can state. The budget questions we suggest before you approve next year’s technology spend are a good place to begin.
Then name one executive who owns the answer, with a date for the first report. It should not be the IT manager alone, because the decisions are about customers, staff and money, and those are the CEO’s. Picture a distributor whose quote desk costs $1.5 million a year in payroll. A 20 percent improvement is $300,000 a year, an illustrative figure, and it would pay for a great deal of careful testing. A company that can name its quote desk, its starting cost and its owner is ahead of many companies that use more tools.
What This Will Not Do
It will not tell you how your industry compares. The Census tables break results out by sector, and the Minneapolis review shows the spread is wide, so check your own. It will not tell you whether AI will pay off at your company. The Census question is self-reported, measures any use at all, and covers two weeks, and the size groups are headcount bands, not revenue bands. Our own list of what to tell the board is a judgment drawn from our work with mid-market companies, not a measured standard. We have also not tested it against every industry. If you mention the state hiring laws in your board materials, remember that this is not legal advice, and ask your counsel which laws apply to you.
Where to Start
Before your next board meeting, send your leadership team three questions in writing and give them a week. Where do we use AI today, officially and unofficially? Which business measure has moved because of it, from what starting number? Who owns it? If the answers come back vague, you have your answer to “are we behind,” and you got it before a director asked.
If you want an outside read, our technology assessment gives you a plain-language picture of where you stand, and CIO IQ gives your leadership team an experienced CIO’s advice at $1k per week. You can also reach us through the contact form on our about page.


