The Daily Dig
Construction's productivity problem has been well documented for years, but a new McKinsey analysis puts fresh numbers behind how AI could start closing the gap. The firm projects AI and automation could unlock roughly $228 billion in annual value for the U.S. AEC industry by 2030. A separate estimate puts the impact on Europe's construction sector alone at roughly $126 billion.
The backdrop explains the urgency. Global construction output hit about $15 trillion in 2025, and demand could reach $22 trillion by 2040. Yet construction productivity grew just 10 percent between 2000 and 2022, compared to 90 percent in manufacturing over the same period. Left unaddressed, McKinsey estimates that gap could leave construction output short of demand by up to $40 trillion cumulatively by 2040.
The report is direct about what this means for the workforce. McKinsey estimates AI could automate 50 percent of nonphysical work in architecture and engineering, and 39 percent in construction. That spans more than 150 workflows across 25 domains, from bid and pricing analysis to RFI triage to back-office functions like invoicing and document control. Adoption won't be all or nothing. McKinsey advises firms to sort workflows into three buckets: those best led by people with agent support, those best led by agents with human oversight on critical calls, and those ready for full automation.
McKinsey lays out the shift across three time horizons. In the next 18 months, the priority is streamlining fragmented workflows. The report advises firms to prioritize three to five high-value workflows, citing estimating, constructability review, and schedule risk as examples where cost, schedule, risk, and performance variability run highest. Over 18 to 48 months, the advantage shifts to firms that structure and retain their own project data instead of letting archives of drawings, RFIs, and closeout reports sit unused. Beyond four years, McKinsey expects AI to move into site-level execution, including autonomous construction equipment, yard logistics, optimized haulage, and coordination between factories, yards, and job sites. Large-scale robotics and humanoid labor for tasks like welding and plumbing is described as roughly a decade out.
The report also addresses the commercial side. As AI reduces the labor hours behind a given scope of work, McKinsey argues firms need to shift from billing by time toward outcome-based pricing, including fixed fees, milestone payments, or shared-savings arrangements. Without that shift, firms risk giving away their own productivity gains to clients. The report also flags a workforce concern: many of the tasks AI is automating first are the same ones junior staff have traditionally used to build judgment and experience. That could thin out the industry's training pipeline if firms don't adjust how they develop younger talent.
Snapshot:
Projected AI value, U.S. AEC industry (by 2030): ~$228 billion annually
Projected AI impact, European construction sector: ~$126 billion
Global construction output (2025): ~$15 trillion
Projected global construction demand (2040): ~$22 trillion
Construction productivity growth (2000 to 2022): 10% (0.4% annually)
Manufacturing productivity growth (2000 to 2022): 90% (3.0% annually)
Potential construction output shortfall (cumulative, by 2040): up to $40 trillion
Nonphysical work automatable in architecture and engineering: 50%
Nonphysical work automatable in construction: 39%
Workflows identified across AEC domains: 150+ workflows, 25 domains
AI adoption horizons: Near term (0 to 18 months), Medium term (18 to 48 months), Long term (4+ years)
Named example workflows for near-term prioritization: estimating, constructability review, schedule risk
Large-scale robotics and humanoid adoption in construction: estimated roughly a decade away
TheJobWalk Thoughts
McKinsey points to estimating, constructability review, and schedule risk as examples of high-value starting points for AI adoption, areas where cost, schedule, risk, and variability run highest. That's a useful filter for GCs and subs deciding where to spend limited implementation budget first. The report also advises prioritizing workflows where performance today leans heavily on a small number of experienced people. In our view, these three examples fit that pattern well, since estimating, constructability judgment, and schedule risk are exactly where a project typically loses the most when the one person who has seen this before is unavailable.
The bigger shift is commercial. If AI cuts the labor hours behind a proposal or a schedule, firms still billing hourly or by traditional markup risk handing their own productivity gains to clients by default. Sales and BD teams should treat outcome-based pricing as a conversation to start now, not one to wait on until clients raise it first.
There's also a real training gap forming. The report warns that AI is automating many of the same tasks junior staff have used to build judgment. Firms that don't build a deliberate alternative path risk a generation of estimators and project engineers who can run the software but haven't developed the pattern recognition the job eventually demands.
