Will AI Replace Construction Workers? What Modular Home Factories Mean for the Trades

A contractor in Washington State building modular homes. A friend in Toronto doing warehouse work for a company backed by Canada’s largest homebuilder. Two unrelated conversations in the same week, both pointing at the same thing.

Peter Gilgan, who founded Mattamy Homes in 1978, is behind Stelumar Advanced Manufacturing, a fully automated 450,000-square-foot facility in the Greater Toronto Area. The stated target is around 3,000 housing units a year. Mattamy is the first customer. Build time on a unit is estimated at months instead of years, and framing that takes six weeks on site is supposed to take about a day in the plant.

That is public information. You can go read it yourself. What nobody has mapped out is what it does to the guys who actually do the work.

The labour math contractors should run for themselves

Think about how many trades touch a single new build. Framers, plumbers, electricians, HVAC, drywall, tapers, painters, tile, flooring, trim, insulation. The estimate that came out of one of those conversations was roughly $4,000 saved per trade, per house, once the factory is running properly.

Call it fifteen trades. That is $60,000 a house. Multiply by 3,000 units a year and you get $180 million a year in labour that stops flowing to trades businesses.

That number is an estimate, not a spreadsheet from Mattamy. The scale of it is what matters. Even if it is half that, it is still the kind of money that reorganizes an industry.

Will AI replace construction workers, or just move the work

The honest answer is that it replaces a specific kind of construction work, and the timeline is longer than the headlines suggest.

Repetitive, high-volume, production new-build work is what moves into a factory. Framing the same floor plan for the 200th time in a subdivision is exactly what robotics handles well. Custom work, renovation, retrofit, and anything involving an existing structure stays where it is, because you cannot lift a 40-year-old bungalow onto a flatbed and send it back to the plant for a bathroom reno.

There is also a quality argument here that runs opposite to what most tradespeople expect. Anyone who has walked a new subdivision home knows what the workmanship looks like behind the walls when a builder is banging out 300 units with five floor plans. A robotic line that does the same operation with the same tolerance every time is likely to beat that, not lose to it. Weather-independent, less waste, tighter envelope. Precision is what machines are good at.

On the timeline, the realistic floor is around ten years before this is producing at a scale that genuinely displaces trade labour across a market. Factory construction alone runs about a year after land, zoning, and permits. Then comes the ramp, the process tuning, the inspection frameworks, and the union pushback. None of that stops the trend. It stretches it.

The counterweight is government money. Modular housing is a plank in the federal plan to dramatically increase housing starts, with $25 billion in debt and $1 billion in equity earmarked for companies building prefab homes. When the state is subsidizing the transition, the transition speeds up.

The savings are not going to the homebuyer

Here is where most contractors get the story wrong. They assume factory efficiency means cheaper houses.

Housing is priced at what the market will bear. A builder who takes $60,000 of cost out of a unit has a choice: pass it to the buyer, or keep it and use it to outbid everyone else for land. Land is the constraint in every major market. The builder keeps the margin, buys the better lot, and grows share.

The second-order effect is what should worry you. Once the factory builder can undercut the site-built developer across the street by even a little, the site-built developer’s margin gets squeezed. The trades working that site get asked to sharpen the pencil. Then that developer’s land gets bought by the factory builder. Everybody downstream gets compressed.

And this is before a company with existing warehouse infrastructure and capital looks at the model and decides to enter. Once the robotics and the floor space exist, it is press and go.

Which trades get squeezed first

New-build dependent trades feel it first. Framing, drywall, rough-in plumbing, rough-in electrical, HVAC rough-in. If a large share of your revenue comes from builder contracts on production subdivisions, you are the most exposed position in the industry.

The trades that feel safe should look harder. Excavation seems untouchable until autonomous machines handle a straightforward dig. Foundation work seems untouchable until you drop a programmed machine inside pinned corners and let it pour footings and walls with a waterproofing admixture already in the mix. Paving a 12-by-20 driveway is a geometry problem a machine can solve. Roofing already has sprayed coating products that let one person on a ladder set a machine running and come back in four hours.

The pattern is consistent. Any task that is repetitive, spec-driven, and performed on a clean site with known dimensions is a candidate. Any task that requires diagnosing something unpredictable in an occupied building is not.

Service work is where the trades survive

There are roughly 160 million existing homes in North America. Around 1.5 million new ones go up each year. New construction is about 1% of the housing stock annually.

That other 99% does not care who built it or how. Panels still need upgrading. Furnaces still fail. Roofs still leak. Foundations still crack. A tile still breaks, drywall still gets a hole in it, and a faucet still drips at 11pm on a Sunday.

Service work also held up through COVID when almost nothing else did. It is about as close to recession-proof as construction gets, because the demand comes from things breaking rather than from consumer confidence.

If the factory scenario plays out the way it looks like it will, a trades business with a real service arm has a floor under it. A trades business that depends entirely on builder contracts does not. That is the single most useful takeaway from this whole conversation, and it is actionable now, not in ten years.

Building a service arm is an operational problem more than a market problem. You need recurring maintenance agreements, a schedule that holds, technicians dispatched without someone playing phone traffic controller, and invoices that actually go out the same week the work happens. Jobtable is built to run that side of the business without a month of setup.

Learn more at https://www.jobtable.com

One more angle worth watching

If factory-built homes end up carrying integrated systems and proprietary components, the service relationship could get controlled the way car manufacturers control dealer networks. Trades become authorized service providers for the builder rather than independent operators. That is speculation, but it is the kind of speculation worth positioning against early, because the businesses that already have direct homeowner relationships and service contracts in place are harder to disintermediate.

FAQ

Will AI replace construction workers?
Not across the board. Repetitive new-build tasks performed in a controlled factory setting are the exposed category. Service, repair, retrofit, and diagnostic work on existing buildings stays with people for the foreseeable future because every job is different and the conditions are unpredictable.

Which construction trades are most at risk from automation?
Trades tied to high-volume production new builds: framing, drywall, and rough-in plumbing, electrical, and HVAC. The risk is concentrated in builder contract work, not in residential service.

Are modular homes lower quality than site-built homes?
Factory-built units are produced in controlled conditions with consistent tolerances and no weather exposure, which generally produces tighter, more consistent results than volume subdivision building. Quality concerns are more about design repetition than about construction defects.

How long until modular factories affect trades work?
Roughly ten years before production reaches a scale that materially reduces trade labour demand in a given market, factoring in factory construction, process ramp-up, inspection frameworks, and regulatory friction.

How should a contractor prepare for automation in construction?

Reduce dependence on builder contracts and build a residential service and maintenance arm with recurring revenue. That work is tied to the existing housing stock, which is roughly 99% of the market and grows every year.