Statistics Canada added a question this spring about whether using AI changed how a business trains or staffs. Sorted by company size, the answers separate much further than adoption does. Combining the two published rates, we estimate that about 18.9% of Canadian firms with 100 or more employees have both adopted AI and trained employees on it, against about 4.8% of firms with 1 to 4 employees.
Nearly one in five Canadian businesses now uses artificial intelligence. In the second quarter of 2026, 19.2% reported using AI to produce goods or deliver services in the previous 12 months, a share Statistics Canada describes as triple the 6.1% recorded in the second quarter of 2024.
That figure has been reported widely. The one underneath it has not. This year the Canadian Survey on Business Conditions asked whether AI use led to changes in training or staffing practices, and the answers break by company size in a way the adoption number hides.
Adoption is close between small and large firms
Among businesses with 1 to 4 employees, 19.9% reported using AI in the last 12 months. Among businesses with 100 or more employees, the figure was 27.8%.
The smallest firms sit 7.9 percentage points behind the largest on whether AI is in the building at all. On that measure the distance is real but narrow.
Training separates them much further
Of the businesses that reported using AI, 44.4% made changes in training or staffing practices because of it. Split by size, that average covers two very different situations.
Among businesses with 100 or more employees that used AI, 68.1% reported training existing employees and 51.7% reported training existing executives. Among businesses with 1 to 4 employees that used AI, 24.0% trained existing employees and 15.6% trained existing executives.
These training figures are not directly comparable to the adoption figures, because they rest on a different base. The adoption percentages describe every business in a size band. The training percentages describe only the businesses in that band that already use AI. Reading one against the other without adjusting would overstate the distance.
Putting both figures on the same base
The two rates can be combined. If 19.9% of the smallest firms use AI, and 24.0% of those trained employees, then roughly 4.8% of all firms with 1 to 4 employees have done both. Applying the same arithmetic to the largest firms gives about 18.9%.
Statistics Canada does not publish that cross tabulation. The bottom row of the table below is our calculation from the two published rates, and it is an estimate rather than a measured figure.
| Measure | 1 to 4 employees | 100 or more employees | Gap, percentage points |
|---|---|---|---|
| Used AI in the last 12 months (all firms in the band) | 19.9% | 27.8% | 7.9 |
| Trained existing employees (of the firms using AI) | 24.0% | 68.1% | 44.1 |
| Trained existing executives (of the firms using AI) | 15.6% | 51.7% | 36.1 |
| Used AI and trained employees (all firms in the band, Raw POV estimate) | 4.8% | 18.9% | 14.2 |
On that common base, a firm with 100 or more employees is about 3.96 times as likely as a firm with 1 to 4 employees to have adopted AI and trained employees on it. For executives the same calculation gives about 4.6 times. Adoption alone separates the two groups by a factor of 1.40.
A headline number moving in one direction can sit on top of a mix moving in another. We found that pattern in the youth justice figures too, where charges in Canada fell 24% since 2019 while the composition of those charges shifted. The AI figures behave the same way. The share of businesses using it climbs, and the question of who was taught anything sits underneath, moving at a different speed.
Cost is not the barrier businesses name
When the survey asked what limits AI use, the most common answer was not money. Two in five businesses, 40.0%, said AI is not relevant to the business. Cybersecurity or privacy concerns came next, reported by 13.4%. Cost was the second leading barrier at 10.6%.
So roughly one business in ten points at price. Four in ten point at relevance, which is a different problem with a different fix. A firm that has decided AI does not apply to what it does will not be moved by a cheaper subscription. It might be moved by seeing the work done, which is a training and demonstration question rather than a procurement one.
Price is often assumed to be the lever in a Canadian business decision, and it is often not the one that moves behaviour. We looked at a version of that question when we read the value meal numbers in our piece on value pricing and Canadian spending, where cheaper options did not by themselves bring back customers who had left the category.
This is also the shape we found looking at why AI pilots stall inside organisations, in our earlier piece on why AI adoption is a culture question before it is a budget question. The Canadian survey now puts national numbers against that argument.
Where AI use concentrates by sector
Use is not spread evenly across the economy. It was highest among businesses in information and cultural industries (42.3%), finance and insurance (40.4%), and professional, scientific and technical services (32.4%). It was lowest in agriculture, forestry, fishing and hunting (4.5%), wholesale trade (7.9%), and construction (9.2%).
The distance between the highest and lowest sectors is 37.8 percentage points, against 7.9 points between the smallest and largest size bands. The survey does not test which factor matters more, and the two are tangled together, since small firms and large firms are not evenly distributed across sectors. What the published figures show is that the sector spread is the wider of the two.
What a firm with fewer than five people can take from this
The practical reading is narrow but usable. If you run a small business and you have started using AI, you are in ordinary company, because about one in five firms your size has. What would make you unusual is training someone, since 76.0% of the smallest firms that use AI did not train existing employees.
What the data cannot show
Four limits are worth stating plainly.
First, this release publishes AI use for businesses with 1 to 4 employees and for businesses with 100 or more, but not for the 5 to 19 or 20 to 99 bands. Everything above compares the two ends. Nothing here shows a steady slope from small to large, and readers should not assume one.
Second, our common base figures are derived, not measured. Multiplying an adoption rate by a training rate assumes the published conditional rate applies as stated, and it carries the sampling error of both figures. Treat 4.8% and 18.9% as close estimates, not exact counts.
Third, the survey records whether training happened, not whether it worked. A business that ran one session and a business that changed how its staff are supervised both answer yes. Nothing here measures hours, quality or outcome.
Fourth, the survey reports what businesses say about themselves. It does not verify any of it.
Frequently asked questions
How many Canadian businesses used AI in the last 12 months?
19.2% of businesses reported using AI to produce goods or deliver services in the 12 months before the second quarter of 2026 survey, which Statistics Canada describes as triple the 6.1% recorded in the second quarter of 2024.
What share of businesses changed training or staffing because of AI?
Among businesses that reported using AI, 44.4% made changes in training or staffing practices. Among those using AI, training existing employees was reported by 68.1% of businesses with 100 or more employees and by 24.0% of businesses with 1 to 4 employees.
Is cost the main barrier to AI use in Canada?
No. The most common barrier reported was that AI is not relevant to the business (40.0%). Cybersecurity or privacy concerns were reported by 13.4%, and cost was the second leading barrier at 10.6%.
Sources and method
From the source. All percentages for AI use, training and staffing changes, barriers and sector breakdowns are published by Statistics Canada in Analysis on artificial intelligence use by businesses in Canada, second quarter of 2026, drawn from the Canadian Survey on Business Conditions, second quarter 2026 (Table 33-10-1167-01). Collection ran from April 1 to May 6, 2026, with a sample of 21,105 and responses from 9,251 businesses or organisations.
Raw POV analysis. The gap column, the estimates of 4.8% and 18.9%, the ratios of 1.40, 3.96 and 4.6, and the comparison between the 37.8 point sector spread and the 7.9 point size spread are our own calculations from the published figures. Statistics Canada publishes neither the gaps nor the cross tabulation of adoption and training. We have flagged in the body which figures are measured and which are derived, and we have not treated the adoption and training percentages as sharing a base where they do not.
Scope. This piece reports survey results about business practice. It does not assess, recommend or compare any specific AI product or provider, and no vendor is named in the figures used here.
