Most of the AI statistics that get quoted at Canadian contractors come from American surveys, often commissioned by companies that sell AI software. This year, Statistics Canada published its own numbers, with an industry breakdown, and construction lands near the bottom of the list.
Why this source is different
Statistics Canada measures AI use through its Canadian Survey on Business Conditions. The headline release appeared in The Daily on May 27, 2026, and a companion analytical article on AI use by businesses in Canada followed on June 11, 2026. The survey drew a sample of 21,105 businesses and produced 9,251 responses, collected in the second quarter of 2026.
Two things make this more useful than most of what gets cited. It's a government survey, so there's no vendor behind it with a reason to make adoption look better or worse. And it's Canadian, so it doesn't require stretching a U.S. sample to fit a Canadian business.
It still has limits, and they matter:
- The construction figure covers the whole sector as Statistics Canada defines it, from large commercial general contractors down to small residential trades. It isn't an owner-operator breakout.
- There's no British Columbia or Okanagan breakout. The only geographic split is urban versus rural.
- Manufacturing doesn't get its own adoption percentage in this release. It shows up only in the data on barriers.
A separate survey from the Canadian Federation of Independent Business puts small-business AI adoption much higher, but it doesn't break out construction or manufacturing, which is why I've stuck with the Statistics Canada figures here.
The national number tripled. Construction didn't keep up.
In the second quarter of 2026, 19.2% of Canadian businesses reported using AI to produce goods or deliver services over the previous 12 months. A year earlier the figure was 12.2%. Two years earlier it was 6.1%. The national rate has roughly tripled in two years.
Construction sits at 9.2%, less than half the national rate. Only two sectors were lower: wholesale trade at 7.9%, and agriculture, forestry, fishing and hunting at 4.5%.
At the other end of the list, information and cultural industries reported 42.3%, finance and insurance 40.4%, and professional, scientific and technical services 32.4%. Information and cultural industries are at more than four times the construction rate.
This isn't an opinion about who's falling behind. It's a federal survey with more than nine thousand business responses, and construction is near the bottom of the list it produced.
If that sounds familiar, it should. Last year's Greater Vancouver Board of Trade report found something similar for B.C., which I wrote about in AI Adoption in BC Construction: Why the Okanagan Is Lagging Vancouver. The new Statistics Canada data is a different survey with a different method, and it points the same way at the national level.
Reading 9.2% carefully
It's worth being precise about what the question measured. Statistics Canada asked whether businesses used AI to produce goods or deliver services over the previous 12 months. That's a specific framing, and it may not capture every way a construction company touches AI. An estimator who uses a chatbot to tidy up a proposal, or an office manager whose accounting software quietly added AI features, may or may not have been counted, depending on how each respondent read the question.
So 9.2% is best read as the share of construction businesses that recognize themselves as using AI in their work, not a precise count of every AI feature running somewhere in the sector. That doesn't weaken the comparison. Every industry answered the same question, and construction answered yes far less often than most.
It also means the gap isn't only about tools. It's partly about whether construction owners see AI as part of how their business operates. In a lot of firms, it isn't on the radar as a business decision at all.
Rural businesses are at half the urban rate
The survey's one geographic split is urban versus rural. Urban businesses reported AI use at 21.0%. Rural businesses reported 9.9%, less than half.
That's not a regional figure for the Okanagan, and I wouldn't treat it as one. Plenty of Okanagan businesses operate in cities, and plenty don't. But it's the closest proxy the data provides, and for a construction or trades business outside a major centre, it suggests two slow-moving categories stacking on top of each other: a low-adoption industry and a low-adoption geography.
Neither disadvantage has much to do with capability. Rural firms and construction firms aren't less able to use these tools. They tend to be further from the vendors, the peer examples and the people who would show them where AI fits. Eighteen years around job sites taught me that being outside the busy market doesn't make the work easier. It usually just means the help arrives later.
"Not relevant" is the most common answer
The most revealing part of the survey is why businesses aren't using AI.
Cost wasn't the top reason. Only 10.6% of businesses cited the cost of using AI as a barrier. Cybersecurity or privacy concerns were the leading specific barrier, at 13.4%. And 23.0% said nothing was limiting their use of AI at all.
The single biggest answer, at 40.0%, was that AI isn't relevant to the goods they produce or the services they deliver.
For some businesses that's probably true. For many others, the conclusion likely came from a quick impression of a chatbot rather than from anyone mapping AI against the estimating, scheduling or paperwork that fills their week. Both reasons land on the same checkbox, and the survey can't tell them apart.
The size breakdown is telling. Among businesses with 1 to 4 employees, 41.4% said AI isn't relevant, compared with 21.3% of businesses with 100 or more employees. Smaller firms are about twice as likely to have concluded it doesn't apply to them.
I made a related argument about last year's GVBOT data in What the GVBOT AI Report Missed: a self-reported survey can tell you what owners believe about AI, but not whether that belief has been tested against their own operation. The new federal numbers make the same gap visible at a national scale.
Testing "not relevant" without buying anything
There's a simple way for an owner to check whether "not relevant" holds up for their business, and it doesn't involve any software. For one ordinary week, keep a rough list of the office work that eats time: quotes written from scratch, emails answering the same questions, change orders typed up after the fact, reports assembled from three different places, time spent looking for a document someone swears they saved.
At the end of the week, look at the list. If almost everything on it requires on-site judgment, physical work or a conversation only you can have, AI may genuinely not be relevant yet, and that's a fine conclusion to reach with evidence behind it. If a large share of it is repetitive writing, sorting, summarizing or searching, then "not relevant" was probably a guess.
That exercise won't tell you which tool to use or where to start. It will tell you whether the question deserves a closer look, which is more than the survey checkbox can.
Manufacturing's barrier is people, not price
Manufacturing doesn't get its own adoption number in this release, but it does appear in the barrier data. Among manufacturers, 12.0% named a lack of skilled workers as a barrier limiting their use of AI.
That's a barrier finding, not an adoption rate, and it shouldn't be read as one. What it shows is that for a meaningful share of manufacturers, the limit isn't the software or its cost. It's whether anyone on staff can run it.
That's a staffing problem showing up as a technology problem. A tool that needs a specialist to operate it full time isn't a fit for a shop that's already short-handed. A tool built to be run by the people already on the floor is a different conversation. The barrier is real. It just isn't the one most people assume.
The gap between big and small is already measurable
Businesses with 100 or more employees reported AI use at 27.8%. Businesses with 1 to 4 employees reported 19.9%.
An eight-point gap might not sound dramatic. But the overall rate tripled in two years, and in a category growing that fast, a gap doesn't stay the same size on its own. Larger firms have more budget, more staff and more room to experiment, and each quarter gives them more experience to build on.
Waiting doesn't hold your position. It hands the head start to whoever didn't wait.
That's not an argument for buying something this quarter. It's an argument for finding out, this quarter, whether "not relevant" is a conclusion you've actually tested.
Why the size gap matters more in construction
Construction in Canada is made up largely of small firms. That means the size gap and the industry gap overlap. A sector dominated by businesses with a handful of employees will tend to show lower adoption simply because smaller businesses everywhere report lower adoption, and are more likely to say AI isn't relevant.
That's worth keeping in mind before reading 9.2% as a judgment on the industry. Part of the gap may reflect firm size rather than anything specific to construction work. But it cuts both ways. If small firms are the ones least likely to have tested whether AI applies to them, then a large share of the construction sector is in exactly that position, and the opportunity, if there is one, sits with the owners who check.
What a construction owner should take from this
A national survey can't tell you what's right for your business. What it can do is put your industry's position in context, with numbers that don't come from someone selling a product.
The context is fairly clear. Canadian construction is well behind the national average on AI use. Rural businesses are behind urban ones. Small businesses are behind large ones, and they're twice as likely to have decided AI doesn't apply to them. For a small construction or trades business outside a big city, all three of those apply at once.
None of that means AI is right for every operation. It means that for a lot of Canadian contractors, the decision not to use it was made without much evidence either way. That's a different situation from having looked properly and concluded it doesn't fit, and it's worth knowing which one you're in.
The paid AI Discovery and Readiness Assessment is built to answer that question for one business. It includes a discovery call and a scored review of your workflows, data and team, and ends in a written report on whether AI is actually relevant to your operation and where.