Innovation bottlenecks in German companies arise because excessive process control, a lack of budgetary authority, and political reluctance toward new technologies block decision-making. Targeted industrial policy that supports entire sectors rather than individual projects, and small teams of highly skilled professionals with genuine freedom to act, would be concrete solutions.
Key Takeaways
- Over-processed companies become paralyzed: Having to get every trip approved individually eats up management time and destroys the deep concentration that software development requires.
- AI doesn’t pay off without a foundation: If you don’t build a unified, machine-readable data foundation, you’re just adding another expensive layer on top of existing data silos that no one has cleaned up in 30 years.
- Gifford Pinchot’s principle “Work underground as long as you can” remains the only practical way to drive innovation at all within over-processed corporations.
- Industrial policy and funding are two different things: Funding distributes money without direction; industrial policy identifies an industry and coordinates the development of an entire supplier infrastructure.
- So far, humanoid robots have failed primarily because of the hand, as the necessary tactile sensors for varying grip forces have not yet been resolved, even though the head and movement are technically largely complete.
Why AI Often Doesn’t Pay Off for Companies
Artificial intelligence doesn’t have a standalone business case that pays for itself. The reason lies in the necessary groundwork: AI needs a foundation. A language model like Anthropic’s must be fed with knowledge, and in addition, all company data must be in a format that this model can read and understand.
It is precisely this groundwork that costs a lot of money and initially yields no direct return. Data silos must be eliminated so that AI can meaningfully access company knowledge. This is not a new demand. It has been made for thirty or forty years, and for thirty years it has been acknowledged without any action being taken.
The greater danger is that companies will simply layer a new AI layer on top of it all instead of cleaning up. Then new complicated processes will emerge again, and the whole endeavor will not be worth it. Anyone who grafts AI onto disorganized data structures is repeating an old mistake in a more expensive form.
Over-processed corporations become paralyzed
Many large companies are so heavily over-processed that they are now becoming paralyzed. This is no longer the result of individual malice but has become a characteristic of the system. What was once considered modern now paralyzes the organization.
The contrast with the past is stark. Gunter Dueck describes how, at IBM, he was able to hire a renowned scientist in a single day: boss informed, works council informed, signed, done. Today, something like that is unthinkable.
The budget worked differently, too. In the past, there was a budget per person per year over which one had authority. Travel expenses, software, social security—everything ran through it without every single expense needing approval. Today, every trip must be approved individually, and every small amount is debated.
The damage is measurable in time. Every fragmented task triggers a new process, and that eats up management time and employee time. No one thinks twice about delaying procedures. So week after week passes with follow-up questions, objections, and renewed coordination.
Deep concentration is underestimated in organizations
Deep concentration while programming collapses with a single interruption, and getting back on track takes time. A simple call to lunch can cost a developer about twenty minutes before they’re back on track.
Dueck compares this to a chess world champion who, in the middle of deep thought, is asked whether he wants meat or fish for lunch today. These personality traits of technical professionals are still not understood in many organizations today. Instead, highly focused people are worn down by a barrage of minor tasks.
Germany has forgotten how to measure itself against the competition
The shame of coming in second has disappeared, and that is precisely the problem. In the past, it was a real insult when another country overtook us. When Japan overtook us in car manufacturing and a Corolla sustained less damage than a Golf, that was a deep thorn in our side that led to a race to catch up.
Today, we let ourselves be told at every turn that Scandinavia, the Netherlands, or Singapore are better, without drawing any motivation from it. For every country we compare ourselves to, we find a reason to defend ourselves: Singapore is a dictatorship, and we mustn’t say anything about China. When it comes to Finland or Sweden, all that’s left is defensiveness.
Yet the positive examples have long been there. Education, unity, determination, energy: other countries demonstrably do this better. The reflex to downplay every comparison prevents us from learning from them.
Subsidies are not the same as industrial policy
The government confuses funding with industrial policy, and that is a costly mistake. With funding, billions are released without it being clear what they’re for. Then programs are put out to tender, a year is spent on the definition, and in the end, a company takes the grant because it fits the funding criteria.
No one asks whether the funded project is a good one. No one wants to see a business case. Whoever falls into the desired industry gets their share, often 20 to 25 percent. The newspaper then reports that one euro of funding has triggered a multiple of that in innovation capital, even though the money was simply taken.
Industrial policy, on the other hand, knows what it’s for. It doesn’t fund a single company, but rather the growth of an entire infrastructure. It identifies a sector and creates the conditions for it to emerge.
| Funding | Industrial Policy | |
|---|---|---|
| Goal | unclear, broadly defined | identified sector |
| Review | fits the criteria, no business case | targeted development |
| Scope | individual companies | entire supply chain |
| Role of the state | distribute funds | create conditions |
Special zones could make innovation possible in Germany
Special zones are a concrete tool of industrial policy, and Germany hardly uses them. China works with innovation zones like Shenzhen, where flight permits are simply approved. There, around a thousand flight routes up to 120 meters in altitude have been defined and cleared, allowing air taxis to be tested and operated between cities with populations in the millions.
Dueck applies this principle to German towns. In Bad Füssing, the healing springs are located about two kilometers outside the town, while guests stay within the town itself. A cordoned-off special zone with park-and-ride facilities on the outside and self-driving cars on the inside could connect the two.
The model scales further. If it works in one place, it can be applied to tourist regions where people no longer have to live right on the beach but can instead be picked up from their guesthouse. Two municipalities have volunteered for such a project. The interest is there; what’s missing is the political decision.
Dueck considers the argument that such technologies only serve the rich to be inconsistent. Anyone who rejects air taxis because only the wealthy use them would have to say the same about Porsche or Mercedes models for China. As an exporting nation, Germany must think of its customers, not its own use.
Why the Hand Is the Key to Success in Robotics
The humanoid robot is nearly complete from a technical standpoint; the bottleneck lies in the hand. The head and AI foundation are considered solved, but the grip remains a challenge. A robot doesn’t automatically know how firmly it needs to grip a bottle to hold it without crushing or dropping it.
Humans solve this effortlessly. They recognize glass or plastic, modulate the pressure, and make corrections through thousands of tactile functions. Integrating this sensor technology into a robotic arm is the real challenge, and it is significant enough that a major corporation dedicated solely to the hand would be worthwhile.
This is precisely where Dueck sees an opportunity for Germany. It requires neither a proprietary AI model nor a complete robot, but rather mechanical engineering and ingenuity—in other words, what German industry considers its strength. The question of why no German corporation has specialized in this remains unanswered.
China Wins Through Coordinated Infrastructure
China’s lead in robotics lies not in individual software but in the coordinated development of an entire supply chain. A robot consists of many individual parts: sensors, actuators, and drives. This requires a supplier industry comparable to the automotive sector.
According to this assessment, American companies are focusing more on pure software, where money flows quickly. China, on the other hand, is building up the entire infrastructure collectively and is now developing its own chips, which are intended to replace established suppliers. These chips are not just a concept; they are already being used in current high-end smartphones.
The pace is the real signal. Within one to two years, proprietary chips—manufactured in Taiwan—will emerge, thereby securing independence from Western suppliers. Companies like Xiaomi build smartphones, home appliances, televisions, and cars, and are now establishing a shared infrastructure to support it all.
Intrapreneurship: Work in secret for as long as you can
Anyone who wants to make a difference within a corporation must push it through, against resistance if necessary. Dueck refers to the principles of Gifford Pinchot from his book on intrapreneurship. The most important ones are:
- Work every day as if you were going to be fired. Treat every day as if your job were on the line.
- Work underground as long as you can. Work behind the scenes for as long as possible before the formal processes kick in.
- Work only with the best people. Hire excellent people and stay away from the rest.
The third point is the one most often forgotten today. Instead of gathering thousands of mediocre employees, Dueck advises starting with ten or twenty outstanding people, poaching them from top companies, paying them well, and then letting them work in peace.
OpenAI did that. Microsoft hired the people and told them they could burn through money however they wanted. They weren’t given a budget at all.
Gunter Dueck
Every corporation knows the opposite model. Innovation sits at the very bottom, with several bosses piled on top of it, and as soon as one of them changes, they have everything recalculated and cut off funding. That is the opposite of a clear “go.”
“Do nothing”: the customer who doesn’t want to change anything
Many optimization efforts fail not because of the technology, but because of the customer who doesn’t want to change anything. Dueck talks about his time working on factory optimization: Despite demonstrable savings of around 20 percent, the big orders didn’t come. A venture capitalist gave him the diagnosis: the main customer simply said “do nothing.”
The same pattern repeats with AI. True optimization requires major leaps: inputting factory throughput times into the system, reconfiguring the system, accepting a higher-level command that reorganizes processes. Very few are willing to do that much work.
This resistance is not new, but it is systematized today. In the past, things could still be worked out on a case-by-case basis. Today, the streamlined bureaucracy stands alongside the “do nothing” approach and makes change even harder.
AI code that no human has ever seen
Current AI models generate code of such high quality that it is immediately usable. Developers in Dueck’s circle report that Anthropic’s new model is so good that the generated code can be used directly without ever having been reviewed by a human.
In this process, entire teams of agents now work together. They simulate the roles within a bank—such as loan officers and tellers—replacing humans one-to-one with agents and having them solve a task collaboratively. In such setups, security vulnerabilities have emerged that were not seen for a long time in purely human work.
Dueck does not share the fear of incomprehensible machine code. His impression from his own experiments: The code is cleanly documented, explained step by step, often more thoroughly than humans would allow themselves.
Humanoids learn faster because they are built like humans
A humanoid robot learns faster than a specialized machine because it can be trained using human data and human processes. Once you’ve built a robot that moves like us, ideally all it takes is asking it to watch once and then do it itself.
This does not apply to specialized machines. Every task, such as harvesting asparagus, would have to be specifically designed and programmed. With humanoids, one can draw on existing data about how humans behave.
Concrete approaches already exist. At a company near Metzingen, employees wear sensor suits and practice movements on the assembly line, which are stored and transferred to the robot as a skill. The vision extends to an app store for humanoid robots, where you can rent a skill like asparagus harvesting for a few weeks.
Automation shifts the fiscal balance
When well-paid jobs are replaced by robots, the balance shifts between those who finance the community and those who are supported by it. Dueck uses the term “net fiscal contribution” to describe this. Anyone who pays more in taxes than they receive from the state is fiscally positive. Anyone who is predominantly paid by the state is negative in a purely fiscal sense.
This calculation is deliberately provocative and says nothing about the social value of a job. A teacher or caregiver fulfills a purpose but is negative in the narrow fiscal sense because their salary comes from the state budget. The state remains sustainable only if the positive contributors pay in more than it consumes.
The trend is moving in the wrong direction. While well-paid jobs are becoming replaceable by humanoids, the government itself pays little attention to AI. Anyone who automates well-paid jobs while simultaneously expanding the fiscally negative side undermines their own financing.
