AI adoption is still uneven across the real economy.
According to Census BTOS data, 21.7% of U.S. businesses reported using AI in July 2026.
That number is meaningful, but the spread underneath it is the bigger story.
Information businesses were at 43.7%. Professional services were at 37.6%. Finance and insurance were at 36.0%. Businesses with 250 or more employees were at 38.8%.
Retail trade was closer to 14%.
Small businesses were much closer to the national average than the top AI-using sectors. Among businesses with 1 to 4 employees, AI usage was 20.8%.
The market often talks about AI like adoption is already everywhere. The Census data shows a more uneven picture.
Knowledge-heavy sectors and larger firms are much further ahead. That makes sense. They have more digital workflows, more software budgets, more internal data, and more people whose work can be changed by AI tools.
The lower-adoption categories are just as important for investors.
If AI usage keeps spreading beyond information, finance, and professional services, the next wave of demand may come from ordinary business software, automation tools, customer service systems, payments, logistics, retail operations, and back-office workflows.
If adoption stalls outside the early sectors, the AI story may stay more concentrated in chips, cloud infrastructure, and a smaller group of software products.
The research question is where AI moves from experimentation to daily business usage.
Right now, the data says the adoption curve is real, but uneven.