You can read my full publication on MoreThanDigital.
Next to “investing in AI” government in the developed world is now telling some version of the same story. “We’ll bring back manufacturing.” “We need to protect our factory jobs.” “Its important to rebuild the industrial base.” It polls well, it sounds responsible, and it is almost entirely wrong – on several levels at once. But that is always the problem with populist policy making – the facts dont matter.
I don’t say this from the cheap seats. I’ve advised almost every major economy in the world, and I’ve been brought in specifically to help with reindustrialisation, FDI, and other measures designed to attract manufacturing back. So I’ve watched this narrative get built from the inside and often I was at the table and fighting for more common sense and for data-driven decisions, but its a hot and very political topic.
So why am I saying this? The story rests on a fundamental misunderstanding of what actually happened to manufacturing – and to industrial policy in general. That misunderstanding is now driving real spending toward a place that doesn’t matter anymore.
So before a single policy is defended, the numbers have to be on the table. Because most of this debate is conducted without them. It runs on feeling, on fear, and on the attention given to whoever screams loudest about lost jobs. But what they’re describing as a failure is, in almost every case, just the natural evolution of an economy — and it happens in every country in the world.
The peak already happened — and it happened a generation ago
Manufacturing as a share of GDP did not collapse recently. It peaked decades ago and has been declining ever since, in every advanced economy, without exception.
Take Germany. Manufacturing value-added peaked at roughly 25% of GDP in 1991 and sits around 19% today. In the United States it peaked far earlier — roughly 28% of GDP in the early 1950s — and is now below 10%. The United Kingdom ran above 25% in the early 1970s and is now near 9%. Japan held one of the strongest industrial bases on the planet and still declined into the high teens.
The data lines up too cleanly to be coincidence:
- Germany: peak ~25% of GDP (1991), now ~19%.
- United States: peak ~28% (early 1950s), now ~10%.
- United Kingdom: peak above 25% (early 1970s), now ~9%.
- China: peak ~32% (2006), now ~25% and falling.
- Globally: manufacturing has fallen from roughly 27% of world GDP in 1970 to under 16% today.
These are not failures of policy. They are the same curve, repeated across every country that industrialised. The decline is structural, not circumstantial. No amount of national willpower has reversed it anywhere — and yet everyone wants to jump on the narrative while conveniently forgetting the longer arc.
Every one of these economies is on the same downslope. The richer and more advanced the economy, the further along the curve it sits. There is no example, anywhere, of a developed economy turning this around. The countries that have “defended” manufacturing best — Germany, South Korea, Japan — have merely declined more slowly. They haven’t reversed the curve. They’ve slowed their descent down it.
So when a politician says Germany or the US lost hundreds of thousands of manufacturing jobs, the honest response is: so did China. Job loss in manufacturing is not a symptom of decline. It is a symptom of productivity. The same thing happened in agriculture — and we correctly understood that as progress.
Agriculture already taught us this lesson
In the 19th century, agriculture employed close to half the workforce in countries like Germany. Today it employs around 1.2% there, similar levels across Western Europe, and contributes under 1% of GDP in most advanced economies. In the United States, one farmer now feeds well over a hundred people — against a small fraction of that a century ago.
Nobody seriously proposes that Germany or the US should rebuild a mass agricultural workforce and start fighting for more field hands and manual crop harvesters. The idea is obviously absurd. We mechanised, we automated, we saturated demand, and the sector shrank to a small, highly productive sliver of the economy. We don’t eat more because there are more farmers. There’s a ceiling on consumption, and once you hit it, automation simply pushes prices — and the sector’s GDP share — down. Nothing magical. No policy required. Just technological evolution doing what it always does: more output for the same input.
I call these the hygiene industries. Don’t misread me — they are necessary. You need basic security, you need basic capacity. But I have to be very clear: agriculture and manufacturing are not where competitive advantage or future growth comes from. The mistake is treating a hygiene industry as if it were a growth industry, and building national strategy around defending it.
Manufacturing is now following agriculture down exactly the same curve, for exactly the same reasons. Sitting at around 15% of global GDP today, it will soon fall below 10% — and keep going.
China is the proof, not the exception
Everyone wants to cast China as the villain and I hear it often that “China is the one stealing all our manufacturing jobs”. If that were true, why is China also losing manufacturing jobs by the millions?
China’s manufacturing share of GDP peaked at around 32% in 2006. By 2024 it had fallen to roughly 25%, and the decline continues. The largest, most aggressive, most heavily subsidised manufacturing power in human history cannot hold its own manufacturing share of GDP. It’s shedding jobs not because it’s losing, but because it’s automating and saturating – precisely as the theory predicts and what is almost like a natural way of industries to evolve.
For anyone who doesn’t know what a modern Chinese factory actually looks like: Xiaomi’s latest “dark factory” produces hundreds of thousands of cars with almost nobody on the floor – essentially lights-out automation with maintenance and cleaning staff. (Just Google “Xiami HyperFactory” and look at the pictures of documentations about it. That is the blueprint for the future of manufacturing EVERYWHERE. Output goes up, employment goes down, and the GDP share keeps falling because automation in a saturated market drives prices down. It will play out here too.
If the country that “won” the manufacturing era is already past its own peak, then the idea that Germany, France, the UK or the US can reverse the curve through subsidy and protection isn’t a strategy. It’s nostalgia with a budget line – economic suicide delivered in subsidised stages. And let’s be honest about the human side too: who actually wants to leave a service job to go back to a manufacturing line?
The economic policy illusion — the paradox underneath it
What fascinates me most is that this fixation on the past doesn’t just waste money. It distorts even the stories we tell ourselves about being modern. There’s a specific version of the trap that catches even sophisticated policymakers.
Europe recently celebrated itself as a semiconductor leader. In a narrow sense, it’s true — Europe is strong in automotive and power chips, the mature nodes used in cars, industrial systems, and appliances. But these are not the nodes that will drive the next industrial revolution. They are not sub-20nm logic. They are the mature end of the market. Valuable, necessary — and structurally another hygiene industry. The understanding is so thin that something built for a mature segment gets celebrated as strategic leadership, while the fact that almost none of the future computing layer is produced in Europe goes unmentioned.
Advanced logic at the leading nodes — CPUs, GPUs, AI accelerators, the compute that will drive the next economy — is concentrated in a handful of players outside Europe. Believing that “we make chips” means “we own the strategic layer of the future” is the same error as believing “we have factories” means “we own the future.”
That’s why I call it the paradox. People understand that semiconductors are important. They don’t understand that old semiconductor technology is a hygiene topic, not a competitiveness one. In a world where governments claim they want to prepare for competitiveness, the question should be about future-industry readiness — not old-industry hygiene.
What this actually means for policy
The honest conclusion is uncomfortable, which is exactly why so few are willing to state it. Backward-looking, populist economic decisions are pushing serious countries toward the brink — because instead of investing in the future, they are spending to delay something that cannot be delayed. And there’s a real opportunity hidden in this: regions like Africa could leapfrog and participate directly in the next industrial revolution, unburdened by the legacy everyone else is trying to protect.
So the question for any government should not be “how do we bring manufacturing back.” That question has no comforting answer — only a populist one. The two real questions are these:
- How fast can we automate our existing industry and manufacturing, to keep it competitive while it shrinks?
- How fast can we build — and own — the infrastructure of the next industrial revolution, before it’s too late to participate at all?
This is the part that should worry policymakers most. Just as the industrial revolution required coal, waterways, rail and energy — and you could not make steel without owning that backbone — the next revolution has its own backbone: advanced compute, AI infrastructure, renewable energy, networks, and the digital and software layer that sits on top of all of it.
Right now most countries are operating on rented infrastructure. The hyperscalers — almost all American, a few other non-European players — own the foundational layer. And it’s very hard to build an economic moat when you’re only a user, never a creator. If governments don’t wake up and build sovereign capacity at this layer, catching up will become structurally impossible — in the same way a country with no coal and no rivers could never have led the steel age.
Let me repeat it, because it matters: the good old days of manufacturing as the engine of national prosperity are gone. They are not coming back, any more than the days when half the country farmed are coming back. We can spend the next decade fighting for the past, or we can build the backbone of the future.
With manufacturing already below 16% of global GDP and falling, the choice shouldn’t be hard to make. Dont support the hygiene industries – build the future economic powerhouse by focusing on what matters in future and not what happened in the past. The past is the worst predictor of the future but a good teacher.
