Places that understand that physical production cannot be abstracted forever will prevail.
For example, we have 100+ drone startups in the United States. But our overall drone production capacity (hammers in Civ) hasn't actually increased. We just have 100 companies buying grey market from Vietnam and Indonesia, many of which came from China originally.
The way the system should work is if you want to do a drone startup, you need to build a drone factory. That's what the money is for.
If the startup fails, maybe the market leader buys the factory for cheap. This is how the automobile industry was in the United States - a bunch of those companies went bust, but the factories were often kept online by the winners.
I think the problem is more nuanced than that. The US was effectively "the world's managers", in the sense that their economic might, entrepreneur culture, and push for globalization resulted in a corporate structure where the ownership and executive levels were US whereas non-critical business domains reflected the local workforce, whether it was the US or not.
This setup worked great while the US dominated the world's economy and influenced their allies and trading partners to actively engage in globalization.
Now that Trump is pushing for isolationism, of course things change.
If I were to tell you a country over five years grew its GDP 5% in 1900, that would mean houses and roads and factories and mines and a whole range of things were built.
In 2020, 5% real GDP growth could be an increase in the value of various services. In fact, you might not need to change the physical world at all to achieve that growth.
Nowadays we're more in conflict, partly just because we can afford to be without risking immediate starvation. Billions get spent on swaying an election this way, more billions get spent on swaying it that way. Dollars spent in both directions count as GDP even though each one has another dollar on the other side pushing in the opposite direction. And it's not just elections, everywhere we're building moats instead of solving problems. It's a very busy way to not go anywhere.
We need to move from scalars to vectors--progress towards some goal--otherwise we're just patting ourselves on the back for working hard to no effect.