That note about manufacturing is very interesting. It is indeed very difficult to make 10s, 100s, or 1,000's of things. Let alone millions.
At my company, we (software group) took a tour of the hardware R&D lab. They have a mechanical lab and a machine shop, which does the initial prototyping. A lot of the mechanical engineers on the R&D side work on basically "dev tools" for the big factories.
An assembly has an intermittent issue with something going out of alignment; they go on site, take measurements, create a CAD model, and take it to the machine shop. Then, they take whatever fixer part to the factory and install it to fix the issue on the line.
There are so many pieces of it -- it's just like software really. The factories are only partially automated, so fixtures and mounts for things need to be ergonomic if a worker is going to place a half-finished part into a fixture 1000s of times per day. Parts have to have an obvious orientation, so it's easy to tell at a glance to know how they go into where they need to go. Every part needs to be designed to actually be manufactured, keeping in mind the limits of the machine that will be creating it, whether CNC or a mold, etc. All this, with a minimum of steps to manufacture the part, using as little material as possible, while looking good and functioning as desired.
I went to school to be a mechanical engineer and hardly any of this stuff was discussed. (It was a big research university to be fair, but still) We spent maybe 6h across my entire degree talking about the way things actually get made. Things like, why use a flat head instead a rounded head on a machine screw in a design? How to design a part to be injection molded? Sheet metal? CNC?
We need to onshore manufacturing as soon as possible, because these are all very difficult skills that require ingenuity, craft, and dedication to acquire. They only can be acquired slowly, over time, through a long career, in an environment where volume manufacturing is happening.