Construction is one of the least digitized major industries in the world — by McKinsey’s Global Institute index, it sits near the very bottom. The cost of that shows up in the numbers: construction labor productivity has grown roughly 1% a year for two decades, while the wider economy grew closer to 3%. Large projects still run, on average, around 20% past schedule and as much as 80% over budget. The work got bigger and more complex; the way it’s managed barely moved.
AI is the first thing in a generation with a real shot at closing that gap, and the industry knows it. But knowing isn’t doing. In a 2025 RICS survey of more than 2,200 professionals, close to half had no AI in place and roughly another third were still stuck in early pilots. Dodge Construction Network’s 2025 contractor data tells the same story — fewer than one in five firms are actually changing how they work. The interest is nearly universal. The adoption isn’t.
The reason isn’t stubbornness, and it isn’t cost. Survey after survey lands on the same barrier: data. Around half of firms name privacy, security, and intellectual-property risk as the thing holding them back — and they’re right to. A bid, an owner agreement, subcontractor pricing, a drawing held under NDA — none of that belongs in a public model you don’t control, where it can leak or quietly train someone else’s tool. The caution is rational. It’s also where most of the industry has decided to stop.
Here’s the part the wait-and-see posture misses: that exact objection is the most solvable one on the list. Closed-loop, file-based workflows keep everything inside an environment the company owns — the AI works over your own files, on your own terms, and nothing is pushed out to a service you don’t govern. This isn’t a fringe view. Advisory firms like RSM already tell construction clients that anything involving client data, financials, or proprietary project information belongs in closed, self-contained tools rather than open public ones. The wall most firms are standing behind has a door in it. Few have walked through it.
And industries don’t wait for the cautious. Innosight’s corporate-longevity research shows the average company now lasts about 15 years on the S&P 500, down from over 30 a generation ago, with a large share of today’s index projected to turn over within the decade. The pattern behind that churn is consistent: incumbents treat a structural shift as optional right up until it isn’t, and by the time the proof is undeniable, the new standard has already been written by whoever moved earlier. Resistance doesn’t stop the evolution — it only decides which side of it a company ends up on.
Construction won’t be the exception. The firms treating closed-loop AI as a careful, in-house engineering problem today are the ones who will set how projects run a decade from now — tighter coordination, cleaner records, faster turnaround, without ever surrendering control of the data. The firms waiting to watch the results will spend that decade adapting to a standard they had no hand in setting. I’d rather help write it than inherit it. That’s the bet the rest of my work is built on.
