A few days ago I read a McKinsey report on Building the system-led factory in textile and apparel manufacturing.
It is not about adding more sewing robots.
It is about moving from a people-led factory to a system-led factory.
Old model
1,000 operators, fabric piled on the floor, a supervisor deciding the next step.
New model
Each garment travels on a smart hanger with its own digital identity — SKU, size, who sewed it, defect history. The system routes it to the next free station with the right skill.
This cannot be bought. It is learned by being the factory for 20 years.
The West outsourced cost… and also learning. Bringing manufacturing back is the right diagnosis, but rebuilding know-how is not the same as rebuilding sheds. McKinsey estimated the US would need on the order of 40% more textile capacity just to onshore.
If we take this to construction — high human component, high variability — the advantage is not with whoever has more robots, but with whoever orchestrates the chaos with process + data + vertical AI embedded in the line.
That is the distinction that made the most sense in the debate: chasing generalized AI is one race; putting boring, specific AI (vision, routing, maintenance) where it already generates ROI is another.
Is system-led manufacturing the real moat?
Source: McKinsey, Building the system-led factory in textile and apparel manufacturing (August 2026). The ~40% US textile capacity figure is from McKinsey analysis on onshoring capacity, May 2026.