> So moving one factory helps reduce the impact of tariffs on the final product from 145% to 135%, and to reduce it to zero you would have to move hundreds of factories, with tiny gains at each step.

This is ignoring two things. First, for the earlier components in the supply chain, it isn't a reduction from 145% to 135%. It's a reduction to 0%, because the inputs are fungible commodity raw materials made in dozens of different countries, so moving the factory removes the entire tariff for the company selling that input to some other company. Which is what the manufacturer (i.e. the component seller) cares about, and they're the one who would be moving the factory.

And second, the later stages of the supply chain tend to contribute the most value. Suppose you haven't moved the factory that makes the capacitors, but you move the factory that makes the phones. The phone costs $500 retail, how much of the contribution was the wholesale price of the capacitors? A dollar? 145% of <1% is tiny.

So you both have a large incentive to move the early stages, because the tariff you avoid is on ~100% of the selling price from the perspective of the company making it, and a large incentive to move the later stages, because the later stages are the largest share of the final price.

> "Just move two hundred factories on a different continent" might as well be "just invent a time machine".

Why is moving 200 factories harder than moving one factory, when there are also 200 companies and they each only have to move one factory?