One Canadian business in five now uses AI. The other four are not wrong.

In the second quarter of 2026, 19.2% of Canadian businesses used artificial intelligence to produce goods or deliver services. Two years earlier the figure was 6.1%. It has tripled, and the pace has not slowed: 12.2% in 2025, 19.2% in 2026.

The gap is not where you would expect

The obvious story is big versus small. It is partly true: 27.8% of businesses with 100 or more employees use AI. But the smallest firms — one to four employees — sit at 19.9%, above the national average and above the 5-to-19 bracket. The very small are not being left behind. The lower middle is.

The sharper divide is geographic. Urban businesses: 21.0%. Rural businesses: 9.9%. A gap of more than two to one, in the same country, under the same rules.

Three sectors are doing the work

Information and cultural industries lead at 42.3%, finance and insurance at 40.4%, professional, scientific and technical services at 32.4%. Outside those three, adoption falls away quickly. If your sector is not on that list, you are not behind your peers — you are with them.

What businesses actually use it for

Data analysis (36.6%), text analysis (34.5%), and virtual agents or chatbots (28.2%). Not autonomous decision-making. Not replacing teams. Reading and summarizing faster than a person can.

The reason to stay out

The most cited obstacle is not cost. It is cybersecurity and privacy: 13.4% of businesses name it, ahead of cost at 10.6%. That is a rational objection, not a technophobic one. A small firm that handles client records and cannot say where its data goes is right to wait.

Adoption is not free of consequence either: 44.4% of AI-using firms changed their hiring or training practices, and 32% put employees through AI training. The technology arrives with an obligation attached.

What we take from it

If you run a small organization outside the three leading sectors, the honest advice is not "adopt AI". It is: find the one repetitive reading-and-writing task that costs you the most hours, price what automating it would actually save, and compare that to what it would cost — including the privacy work. If the number is not obviously positive, wait. Being in the 80.8% is a defensible position in 2026.

That calculation is exactly what our AI and automation audit does, and it is designed to be able to conclude no.