German Cars Lost to Chinese EVs, and AfD's Rise Is Just a Symptom
AfD is not a German problem but a Western one: people feel tomorrow won't be better, and the simultaneous loss of competitiveness in Germany's proudest industries — cars, chemicals, rail — is the real fuel for the far right.
The video won't play here. Listen to the audio instead:
The argument · timestamps estimated from transcript position
AfD is not a German disease, it's a Western disease
Judith Dada argues AfD's rise is not a German phenomenon: the US, France, the Netherlands and the UK all show the same tendency — people feel left behind, feel tomorrow won't be better than today or yesterday, and feel a melancholy for ‘the slower world of the past, where their value in their environment was clearer’. She stresses at the same time that AfD is ‘the ugliest expression’ of this problem — among all far-right parties in Europe it is one of the most extreme, in some respects with regrettable similarities to Nazism. But she refuses to treat it as a fringe right-wing problem: in Saxony-Anhalt more than four in ten people voted for it, which shows the problem is societal and needs drastic political change, plus a vision of the future that can bring people along.
— Judith DadaThe neighbour who votes AfD is actually very friendly
Dada lives on the island of Rügen in Mecklenburg-Vorpommern, and in daily life she has not encountered xenophobia; her neighbours are very friendly to her, her husband, her children and her darker-skinned Nigerian father. But she notices her own fear growing: last weekend her father visited and the two walked around Stralsund, the streets full of AfD campaign posters, the occasional German flag hung in front of an apartment block or a parking lot, and she felt a kind of nausea, wondering ‘whether my father, whether I, am seen as not belonging here, as unwelcome’. She says if she saw a French flag in France she would not feel the same unease; but seeing a German flag in the country where she was born, she asks: is this because they love Germany, or because they are far right.
— Judith DadaGermany's last shred of face has been taken too
Dada says the German model ‘very clearly no longer works’: the economy has long been in stagflation, pillar industries like cars and chemicals have lost competitiveness because of high energy costs, and large numbers of jobs have been cut or moved away. She takes German cars as an example: Germany completely misjudged the relevance of electrification and underestimated Chinese automakers' ability to catch up — it used to be assumed Chinese brands could never have the same prestige as German brands, but now go to markets like South Africa, where German cars once sold at a premium, and you see BYD and a crowd of Chinese brands pouring in, and people no longer care whether it's a BMW or a Mercedes. She says the last remnant of German pride — ‘nobody in China can build a high-end car brand’ — has now been taken away too.
— Judith DadaAmerica's non-tech sectors are just as sick as Germany's
Dada offers a comparison: the US has 150 companies with software revenue above a billion dollars, Europe has only about 30. The Mag 7 together are worth almost half of total global market cap, and economic growth is extremely concentrated in the single sector of tech. But she points out this is not just a European or German problem, it's a technology problem: America's existing non-tech industries are suffering exactly the same malaise as German and European industry, US economic growth is broadly driven by capex going into data centres, by the tech and AI boom, while non-tech sectors grow at the same rate as mature German and European industry. She stresses, however, that this does not remove a fact: in industries dense with technology and R&D innovation, Europe simply does not have as strong an industry as the US.
— Judith DadaEurope is absent from every layer of the AI value chain
Dada breaks the AI value stack into three layers: the chip and infrastructure layer (chips, and the data centres that host them and turn them into training and inference compute); the model layer (the models that actually produce intelligence); and the application layer (using that intelligence to build concrete applications that bring value to people's lives). She says Europe plays too small a role at every layer, and the further down the stack you go, the starker the gap. She argues that without a structural answer to ‘what is American tech, what is Chinese tech’, no amount of transformation effort in Europe will solve this root cause — every other problem compounds and grows out of this root cause.
— Judith Dada80% of the economy doesn't need the strongest model
Dada offers a ‘vibes-based number’: of the total value AI creates, 80% does not depend on the strongest Pareto-frontier model; many more efficient, cheaper, possibly open-source models can do the economically valuable work as long as they fall within the intelligence distribution the task requires. Her company is heavily betting on this direction, namely ‘model routing’ — you don't care whether it's model A, B or C, you only care whether at a given moment it is the cheapest, lowest-latency optimal intelligence. The remaining 20% she admits she finds hard to price: that is where frontier intelligence matters, such as drug discovery, adversarial markets like cybersecurity (there is always a better hacker, so you always want the strongest intelligence to attack or defend), hedge funds, critical infrastructure protection. She uses Einstein as an analogy: give a hedge fund a thousand Einsteins and its success rate rises sharply; give Coca-Cola a thousand Einsteins and it won't make it more successful, because that is brand, a century-old formula, distribution and a physical business.
— Judith DadaTo build frontier yourself, it has to be a Manhattan Project
Dada argues that for the 80%, Europe can get by with the existing model landscape, using Chinese models, American models; but for the 20% it is ‘absolutely critical’ — you cannot honestly tell a European that in ten years you can defend their life and secure their access to the modern medicine of that time if you have not secured access to frontier intelligence. There are only two ways to get access: produce it yourself, or trade for access. If you produce it yourself, being third or fourth is not enough, being ‘DeepSeek minus twelve months’ is not enough, it must be a Manhattan-style effort, extremely ambitious, extremely well funded. She admits Europe is already late on the recursive self-improvement loop, but ‘it's not over until it's over’, and even at the last metre of a marathon she will try not to come last. At the same time she says that even if Europe mobilises capital and talent, success is not guaranteed, so it must plan for failure.
— Judith DadaWhat Europe lacks isn't money, it's people who can make the call
Dada thinks the capital problem is real, but it is not that ‘Europe has no money’ — Europe has pension funds, insurance companies with huge balance sheets, and the European economy is about two-thirds the size of the US. The real problem is a structural lack of the kind of few players the US has: AWS, Google, Microsoft, xAI and OpenAI can coordinate at extremely low cost, xAI can just call OpenAI, Anthropic, Microsoft. Europe has national borders, different regulations, fragmented capital, far higher coordination costs, and not even a unified financial market. Her solution is not to wait for 27 member states but a ‘coalition of the willing’: the largest few companies plus the largest few governments, together pushing pension and insurance money to allocate more to venture capital. She thinks Europe's biggest structural gap is the lack of an Elon Musk figure — someone with enough influence, power and orchestration ability to put all the pieces together, lock people in a room for three days and get the deal done, Manhattan-style.
— Judith DadaIn their own words · checked verbatim
I do think that the AfD is maybe the ugliest expression of this problem, because looking at all the far-right parties across Europe, it is one of the most extreme parties.
Judith Dada0:00
We need to give people the feeling again that the pie is growing for everyone.
Judith Dada0:00
So the last vestige of German pride—that no one will ever be able to have a premium car brand in China—is now being taken from us as well.
Judith Dada0:00
If you look at the tech sector overall, in the United States there are 150 companies that make over a billion dollars in software revenues. In Europe, there are about 30.
Judith Dada0:00
you can be critical of the political views, or the worldviews, or the moral fiber, of these people, but that does not give you the right to ignore what they are saying
Judith Dada0:00
if you produce it yourself, it’s not fine to be number three, it’s not fine to be number four, it’s not fine to just do what DeepSeek does, minus twelve months.
Judith Dada0:00
The only thing that is for certain is that if times change but you refuse to change and remain the same, the result is eventual decline.
Judith Dada0:00
Figures
| AfD support in Saxony-Anhalt | over 40% | 0:00 |
| US companies with software revenue above a billion dollars | 150 | 0:00 |
| European companies with software revenue above a billion dollars | about 30 | 0:00 |
| Mag 7 combined market cap as share of total global market cap | close to half | 0:00 |
| Size of the European economy relative to the US economy | about two-thirds of the US | 0:00 |
| US institutional capital allocated to venture capital relative to Europe | 10 to 20 times | 0:00 |
| Share of total AI value that does not depend on frontier models (Dada's own description: a vibes-based number) | 80% | 0:00 |
| Enterprise productivity unlocked per euro of inference spending | 3, 4, 5 to 10 euros | 0:00 |
Glossary
- model routing
- Not being tied to a single model, dynamically choosing the most suitable model for a task based on price, latency and capability.
- Pareto-frontier model
- The most capable model at the very frontier of the capability-versus-cost trade-off curve.
- Migrationshintergrund
- Official German term meaning a person or their parents have foreign ancestry; Mounk calls it his least favourite German word.
- Aktienrente
- A pension reform under discussion in Germany that would be supported by returns from capital-market investment.
- tech-brosphere
- A pejorative European media term for the Silicon Valley tech scene and its discourse, used to discredit it wholesale.
How to listen
Investors watching European politics and industrial competitiveness, founders building for or expanding into Europe, and engineers who want to understand the geopolitics of the AI value chain.
The opening linguistics discussion about German gender-inclusive language can be skipped; it has little bearing on the main argument.