Few economic topics have received as much attention in recent years as supply chains. From empty shelves to missing semiconductors to blocked canals, there was plenty of material.
Out of it came the thesis that globalisation is reversing and production is coming home. The data supports that only in part, and what's actually happening is more interesting than the simple version.
What the original problem was
The systems that came under pressure had been optimised over decades towards one goal: minimum cost. Just-in-time manufacturing, low inventories, single suppliers per component, concentration of production at few sites.
In normal operation that's highly efficient. It's also extremely fragile, because there are no buffers. If one link fails, the chain stops.
What became visible from 2020 onwards wasn't a single failure but a cascade: demand collapse, then an abrupt shift in demand from services to goods, then capacity bottlenecks in logistics, then catch-up effects that amplified the bottlenecks.
The well-known bullwhip effect — small fluctuations in end demand producing large swings further up the chain — played a considerable role.
What companies actually changed
The rhetoric said reshoring. Reality looks different.
Higher inventories. The most widespread and simplest response. From "just in time" to "just in case." It costs working capital and warehouse space but requires no structural change.
Multiple suppliers instead of one. Dual sourcing for critical components, often across different regions. More expensive to buy, considerably more robust.
Transparency across tiers. Many companies didn't know where their suppliers were buying in turn. Efforts to map this are substantial and not yet complete.
Regionalisation rather than reshoring. This is the central point. Production rarely moves back to Europe or North America; it moves to other countries within the same broad region — closer to the market but still in lower-cost economies.
Why reshoring rarely happens
The reasons are fairly mundane.
Labour cost is only part of the calculation. More important is that entire supplier networks have formed in particular regions — hundreds of specialised firms in geographic proximity, with established processes and available skilled workers.
Those networks can't be replaced by building one factory. You'd have to reconstruct the whole ecosystem, and that takes decades rather than years.
On top of that comes the skills shortage in industrialised countries. Even where firms want to reshore, the workforce is partly missing.
And automation, frequently proposed as the answer, remains difficult for many manufacturing steps. In textile and electronics assembly especially, human fine motor control is hard to replace.
Where genuine change is happening
There are areas where real relocation is occurring, and they share a common feature: political priority plus subsidy.
Semiconductor fabrication is the clearest example. Several major economies have launched substantial support programmes and new capacity is genuinely being built.
Similarly with batteries and certain pharmaceutical precursors. In both cases the motivation is strategic rather than primarily economic.
That's notable, because it represents a return of active industrial policy in regions that had rejected it for decades.
The costs
What's often missing from the debate: more robust supply chains are more expensive, and those costs land somewhere.
Higher inventories tie up capital. Second suppliers offer worse terms because volumes are split. Production in more expensive countries costs more.
Estimates of this effect vary widely, but the direction is clear: part of the price increases of recent years is structural and won't reverse.
That's a legitimate societal choice — more security in exchange for higher prices. It's just rarely stated as such.
What remains
One area where change has been faster than expected is worth flagging: visibility software.
A great many firms discovered during the disruptions that they could not answer basic questions quickly — where a given shipment was, which suppliers depended on a single affected port, what the exposure was if a particular region shut down. Answering those took weeks of phone calls.
Investment in supply chain visibility tools since then has been substantial, and it is one of the less glamorous but more consequential changes. Knowing where things are does not make a chain robust by itself, but it converts a problem you discover when the shipment fails to arrive into one you can see forming three weeks out.
The limitation is that visibility usually extends one or two tiers deep. Beyond that, most companies still cannot see. And the disruptions that hurt most tend to originate further upstream than that — in the supplier of the supplier of a component nobody had thought about.
My impression after several years of watching this: globalisation isn't ending, it's being rebuilt. Less pure cost minimisation, more resilience. Fewer single dependencies, more redundancy. Less intercontinental, more regional.
That's an adjustment, not a reversal. And it's a long way from finished — restructuring of this kind typically takes a decade or more in industry, because it's tied to investment cycles.