It makes no sense for automation to get here before it gets into the developed world. It never worked this way (except when unions made it so) and there's no reason presented for anything to change.
Besides the article spends some time arguing why a level of automation would have a larger impact on developing countries because we rely more on industries and less on services. I think that is too big a focus on the past - we have reasons to think future automation will have a larger impact on services than in the past.
I think the article is interesting, but the conclusions are flawed.
I think you are right, automation will get to the developed world first, but I think in part to automate the developing world remotely.
If you build from scratch, adopting the latest technologies makes sense. If automation technology becomes mature, any new factory will likely use it from the get-go...
Any modern car is going to have an enormous amount of robotic assembly involved. Having humans can give you more flexibility (robots take time to set up), but expose you to more quality problems because they're inconsistent. But many things just can't be done that well by humans.
My first two cars, bought in 1997 and 1999 were pretty prone to breaking and costing a lot of money and time to fix. But the car I bought in 2005 was different. I checked around, and it would seem all brands have suddenly become very fiable. I suppose the factory switched from manual to automated production.
https://outline-prod.imgix.net/20170323-c86vRGKVPsbek3PZXpVw...
Someone else responded with a picture of a car factory in 1984: https://i.kinja-img.com/gawker-media/image/upload/s--OKTGFiU...
They've had automated production in car factories for a long time. And it's not an either-or thing; just like robots were only used for certain jobs back in 1984, and other stuff was done by humans, it's still like that (though probably more stuff is done robotically). Some tasks are just easier to have humans do, or not worth the expense of a robot (and the set-up, jigs, etc. needed for the robot).
Did you buy a different brand car in 2005 than the 97 and 99 cars? Different brands are different. Even within the same brand, there have been improvements over time (and also regressions). Designs have changed; maybe that company focused more on designing for serviceability between those times. Design has little to do with automated vs. manual production. There's nothing stopping you from designing a car to be assembled by hand (like a small production-run handbuilt exotic), and designing it to be easy to change the oil on.
It also depends on what exactly it is that a given country is arbitraging.
I've been considering China for a few years as not selling its labour or resources but its sinks: the willingness of the country to pollute its air, water, and land, in building up wealth and/or capital.
Other opportunities might be space, shipping access, or the ability to bulldoze through any competing property claims. In that case, say, bulldozing a slum in Mumbai to throw up a frabrication plant that can import raw materials and export finished goods from a factory with a very low labour footprint ... could make sense.
I'm not saying it does, but if you're going to bring automation to a place such as India, then you'd look for a dynamic like that: high-quality or low-tolerance manufacturing, probably dirty, with a high transport need, for which the ability to secure seaport-adjacent property with minimal fuss would be expedient.
A hundred fifty years later the (manual) handloom textile industry is still the second largest source of employment in Rural India after agriculture.
So why has automation of the loom not displaced all these weavers? There are many factors and one of the big ones is - automation isn't magic. It doesn't self-fix flaws. It doesn't upgrade itself. Master weavers can.
The looms weren't AI -- just automation.
In any environment other than grinding poverty, automation is cheaper than human labour.