Most of the talk about AI in construction is about software: estimating tools, scheduling assistants, document review. A survey released this month points to a different route by which the AI boom is already reaching contractors. Not through a tool anyone bought, but through the hiring market.

Every AI data centre going up has to be built by somebody. Electricians, pipefitters, HVAC technicians. The same people smaller contractors have been trying to hire for years.

The survey, and its limits

The Associated General Contractors of America and NCCER released their 2026 Workforce Survey analysis on September 3, 2026. The fieldwork ran in July and August 2026, and 1,830 people from a broad range of firm types and sizes responded to at least part of it. It's the first year the survey directly measured how data-centre construction and AI are affecting the construction workforce. Engineering News-Record and others covered the release.

Three caveats.

It's American data. There's no B.C. or Okanagan breakout, and I'm not aware of a Canadian survey that measures the same data-centre-versus-local-trades competition. The mechanism isn't specific to the U.S.: a fast-growing, well-funded construction segment bidding up a limited pool of skilled trades. But the percentages belong to U.S. contractors.

It isn't a small-shop sample. Among firms that reported revenue, 58% did $50 million or less of work in the previous 12 months, 33% did between $50.1 million and $500 million, and 10% did more. That's broader than a big-contractor survey, but it isn't a survey of owner-operators either.

The sponsors have an interest. AGC is a trade association that lobbies for workforce funding and immigration pathways, and NCCER is a training and credentialing body. Both have institutional reasons for shortages to read as serious. The method, sample and fielding window are disclosed and respondents weren't paid, but it isn't a disinterested source.

One thing I've deliberately left out: several secondary write-ups attach large national shortage totals to this story. Those numbers don't come from this survey, and I couldn't trace them to a single primary method, so they aren't used here.

The bidding war you never entered

Twenty-eight percent of respondents did construction work on a data-centre project in the past 12 months. Among firms that reported on how those projects affected their workforce:

Those percentages are of firms reporting on those effects, not of all 1,830 respondents. That qualifier matters, and it's how AGC presents them.

The point is still blunt. A contractor who never bid on a data centre is competing with one anyway, for the same crew, in the same labour market, against a budget that looks nothing like theirs. The AI boom reached that business through an electrician's paycheque before it reached it through any software.

Posting the job and filling it are different things

I spent 18 years in construction before this work. The trade everyone fought hardest over, every year, was electricians.

That hasn't changed. It's gotten worse. Of firms with electrician openings, 81% report difficulty filling them. Mechanics follow at 79%, HVAC technicians at 77% and concrete workers at 76%.

Zoom out and the picture is similar across the board. Eighty-seven percent of firms have openings for hourly craft workers. Among those, 88% say the positions are as hard or harder to fill than a year ago, and 50% say they're harder.

It isn't only the field. The same survey found 72% of firms with estimating openings report difficulty filling them. The office roles that keep work flowing into the business are short too.

A growth plan that starts with "once we get another person" isn't really a plan under those conditions. It's a waiting list.

A slowdown that didn't help

Normally a softer market loosens up labour. People come available, wages settle. That isn't what this survey shows.

Thirty-seven percent of firms cut their headcount by at least 5% over the past year, against 34% that grew it by at least 5%. So the market genuinely softened. Yet 73% expect to add employees over the next 12 months, and only 8% expect to reduce headcount.

The more telling detail is who grew. Across all firms, 34% increased headcount. Among firms with $50.1 million to $500 million in revenue, that rose to 54%. Among firms over $500 million, it was 79%.

A soft market didn't level the field. It sorted it by size.

If you're on the smaller side of that split, outbidding a firm ten times your size for the same tradesperson isn't a realistic strategy. The leverage has to come from somewhere else.

Raises bought a tie, not a win

Contractors are spending to compete. Fifty-five percent of firms increased base pay for hourly craft workers by more than they did the year before. That's real money out the door.

It didn't close the gap. Half of firms still say the candidates available lack the skills, certifications or licences needed. Forty-two percent say new hires don't show up or quit shortly after starting. And 83% report at least some turnover among new field employees within their first 90 days.

Those numbers point somewhere specific. If pay were the binding constraint, raises would fix it. They aren't fixing it, which suggests the problem is about supply and retention more than price. You can't pay your way to a qualified electrician who doesn't exist in your market, or keep a new hire who leaves in week three because the job wasn't what they expected.

That changes the operating question. Not "how do I pay enough to win the hire," but "how much of this work actually needs that hire?"

Which work that question applies to

This isn't an argument that software replaces tradespeople. It doesn't. Nobody's pulling wire with a language model. But a skilled person's week in a small contracting business is rarely all skilled work. It includes quoting, paperwork, chasing information, re-explaining job details and writing up what happened on site. Every hour of that is an hour the scarcest person in the business isn't doing the thing you can't hire anyone else to do.

Picture a twelve-person electrical contractor

Take a hypothetical electrical contractor with a dozen people. The owner is also the main estimator and the person clients call. Two senior electricians run the crews, and one of them is the only person who really understands how the company prices service upgrades. The office has one person handling payables, scheduling and permits.

Now look at where the skilled hours go. The owner spends evenings writing quotes from scratch. The senior electrician spends part of each week answering pricing questions from the office and rewriting job notes so the next crew can follow them. The office manager retypes information that already exists in three places. None of that is wiring. All of it competes with the work only those people can do.

If that business could hire another licensed electrician tomorrow, it would. Based on the numbers above, it probably can't. What it can do is look hard at the non-trade hours its scarcest people are carrying, and decide which of those hours could be handled some other way. That's where technology, including AI, has a realistic role in a labour shortage: not replacing the trade, but protecting the time of the people who have it.

The 90-day problem

The retention numbers deserve their own note. AGC reports that the reason contractors most often give for new field employees leaving early is a mismatch between workers' expectations and the realities of construction work. That's an onboarding and communication problem as much as a hiring one. Clear job descriptions, consistent orientation material and written procedures that a new hire can actually follow are all things a small firm can improve without a big budget, and they're also the kind of documents that tend to get written once and never updated.

What contractors actually expect from AI

The survey asked contractors about AI and robotics directly, and the answers aren't what the headlines about job losses would suggest.

Fifty-one percent expect AI and robotics to have a somewhat or very positive impact on construction over the next five years. Eight percent expect a negative one. That isn't an industry bracing for replacement.

Asked where AI would have the biggest impact inside their own firms, contractors named:

Office work. The hours that pile up after the crew goes home. These are expectations about the future, not measured results, but they line up with where the labour shortage bites hardest for the people running the business.

Meanwhile, 42% said shortages of their own workers or their subcontractors' workers have already delayed projects, the most commonly cited cause of delay in the survey.

Expectations are not results

It's worth separating what this survey measured from what it didn't. The 67% figure for estimating and bidding is where contractors expect AI to matter most over the coming years. It isn't a measure of where they use it today, or where it has already produced a result. Expectations in surveys like this tend to run ahead of reality, sometimes by a long way.

That doesn't make the expectation meaningless. When two-thirds of contractors point at the same part of the business, it says something about where they feel the pressure. Estimating is skilled work, the people who do it well are hard to replace, and it's also one of the roles the survey found hard to fill. Contractors aren't hoping AI will run their crews. They're hoping it will take some of the load off the few people in the office who keep work coming in.

What this means north of the border

To be straight about it again: these are U.S. percentages. The pressure they describe doesn't stop at the border, and B.C. contractors already know what it's like to compete with large projects for the same trades. But I wouldn't quote any of these numbers as a description of the Okanagan.

What does carry over is the logic. When skilled people are scarce and expensive, the business that gets more output from the crew it already has is in a stronger position than the one waiting to hire. That usually starts with finding the non-trade work your trades and your office are carrying, and deciding which of it could be handled differently.

Sorting which jobs in your business genuinely need another set of hands, and which are processes nobody has rebuilt, is what you're paying for in the AI Discovery and Readiness Assessment. It's a paid engagement, not a free consult, and it ends with a written report on where your existing team's time is going and where the leverage is.

Get details on the assessment →