This might be exaggerating a little bit, but I do think that repairing stuff is inherently more difficult to automate than production, and increased automation is the major way we've seen productivity gains in manufacturing other than offshoring to cheaper countries and shipping things over slowly but cheaply in bulk.
Unlike in assembly, there's not just one happy assembly path that you need to handle – first you always need to diagnose the problem, and then depending on the problem take a variety of actions in order to fix/replace whatever you've identified as the faulty component.
Plus subsequently, even in the ideal case of a widget that's as easy to disassemble/repair as it was to originally assemble (which comes with its own set of trade-offs), any productivity gains in manufacturing that were based on parallelisation are difficult replicate when repairing just a single one-off component. E.g. reflow soldering allows you to efficiently solder large circuit boards with hundreds or more components in one go, but conversely subsequently replacing just a single one of those components therefore doesn't take one hundredth of the original assembly time.
that's false. We can also make repairs materially faster (more productive).
Just to illustrate the point with something extreme. If a worker could fix 100 fridges a day, they'd only have to charge like $5 labor.