The most useful thing in McKinsey’s new AI report is a number that cuts against the hype. Artificial intelligence isn’t an extinction event for design and construction firms, the firm argues, but it could automate 39% of the sector’s nonphysical work, and half of it in architecture and engineering.
A map, not a prophecy
The July 15 report identifies 150 workflows across 25 AEC domains, each with a different ceiling for automation. Data entry, invoicing and equipment inspection are set to change most by 2030. McKinsey sorts the opportunity into three windows: the first 18 months on office tasks like bid/no-bid analysis, estimating and proposals; the next few years on turning proprietary data such as RFIs, drawings and closeout reports into an edge; and, past four years, autonomous equipment and logistics on the jobsite itself.
Daniel Ahmoye, a McKinsey partner, made the sharper point in an interview: early wins in design and modeling ‘will likely soon be table stakes.’ The advantage goes to firms that redesign how they work fastest, not the ones that bolt a chatbot onto old processes.
Build, buy, or neither
The report wades into the debate high-profile builders have already joined, with firms like Suffolk building tools in-house. McKinsey’s counsel is blunt: AEC firms have struggled to build and scale software, and AI moves too fast for most to rely on internal development. Build where your expertise is the product, the firm says, and buy where an outside vendor is investing more than you ever could. Framed that way, it stops being a build-or-buy question and becomes a simpler one: where does our advantage actually come from? For context, global construction productivity improved just 10% from 2000 to 2022, so the bar to clear is low, and the prize for clearing it is large.