Regulation vs. Speed of Technology: A Structural Tension Field

At the outset, the question was discussed whether political processes can keep up with the rapid development of AI at all.
Khubbueva puts it clearly:

“The simple answer is no.”

In the original AI Act, it was already evident that the attempt to anticipate future technological developments hits real limits. Thresholds for generative AI were originally supposed to regulate large US models but often also affect German industrial companies in practice.

The AI Act is currently being reopened – partially fundamentally. The goal: to create legal certainty.

“Legal certainty is a form of competitiveness.”

Companies need clear conditions before they invest in AI applications, compliance, or safety-critical processes.

Regulation as a Factor – But Not the Sole Reason for Innovation Problems

The narrative is often served that Europe is overregulated and thus hostile to innovation.
Khubbueva counters a simplified representation:

• Regulatory overlaps – such as between the AI Act, GDPR, Machinery Directive, or Medical Devices Regulation – are real and can stifle innovation.

• At the same time, structural challenges exist that have nothing to do with AI regulation:

  • a fragmented internal market

  • missing capital market union

  • high energy prices

  • a complex investment environment

Companies with heavily regulated products feel the burden particularly: longer time to market, high compliance costs, and uncertainty in interpretation.

Why Global AI Governance is Rarely Realistic

The discussion also shed light on the international perspective. In recent years, there have been several initiatives for global AI governance – from the Hiroshima process to the AI Safety Summit in London.

However, the course change in the USA and China's increasingly strategic turn towards AI application complicate genuine international agreements.

“Currently, one can only agree on very fundamental things.”

While Europe prioritizes governance, China strongly focuses on industrial AI application – and invests massive amounts. This is shown, among other things, by the fact that China has already overtaken Germany in the automation of production facilities.

Europe Weak in AI Development – Strong in AI Application

A central point of the discussion was the actual status of Europe in the AI sector:

• Germany: only 2% share of global AI development

• No major language model manufacturers left in Germany

• In Europe, one player dominates: Mistral

Nevertheless, there is enormous potential:

“Germany has one of the strongest industrial bases in the world.”

Strengths exist particularly in:

• machine engineering

• automotive

• energy

• chemistry

• medical technology

• high-quality industrial data

• robotics and physical AI

• research and education

Here, Europe – and particularly Germany – could still be globally leading in the future.

The Role of Language Models – And Why Europe Must Differentiate

Khubbueva emphasizes that Europe does not necessarily have to compete in the race for the largest models.
Small and mid-sized language models are realistic, meaningful, and necessary, especially for industry-relevant applications.

The challenge lies less in technical ability, but in:

• missing investment volumes

• fragmented funding landscape

• lack of scaling

Europe produces many good ideas – but too few that can be scaled globally.

Regional Support is Valuable – Scaling is Decisive

It became clear in the discussion that regional and national funding programs for startups and SMEs are extremely important.
However, Khubbueva points out a structural problem:

Funding programs create many pilot projects, but rarely create scalable ecosystems.

The ideal would be:

• regional support for early phases

• European instruments for scaling

• clear industrial policy priorities

AI Act: Why the Reopening is an Opportunity

The renegotiation of the AI Act offers one thing for the industry above all:

• more time

• more clarity

• better alignment with existing laws

The BDI demands a postponement of deadlines – originally 24 months, currently under discussion: 16 months.

Particularly important is the interplay between GDPR and AI Act, especially regarding data use for training AI.

Where Regulation Actually Helps

Regulation is not per se an obstacle to innovation.
In areas such as chemistry or dangerous machinery, it creates safety and clear processes that, in turn, promote innovations.

It becomes problematic when new regulations create duplicate compliance structures, are contradictory, or tie up resources that companies would prefer to invest in AI development or cybersecurity.

Semi-Conductors: Why Chips, AI, and Industrial Policy are Inextricably Linked

Khubbueva vividly explains how fragile and geopolitically charged global chip supply chains are.
The case of Nexperia showed how political decisions can suddenly jeopardize entire value chains.

“With chips, it’s not just about the economy – it’s about national security.”

Europe must realistically assess dependencies:

• some segments can be sensibly produced in Europe

• others remain permanently globalized

• new partnerships (India, Indonesia, Kenya) are necessary

It is important to focus on European strengths – such as in geometry technologies or highly specialized manufacturing steps.

Digital Sovereignty Needs a Clear Definition

Khubbueva calls for the term “digital sovereignty” to be precisely clarified at last:

• What must be truly independent?

• Where are alternatives and diversification sufficient?

• Where can Europe create chokepoints itself?

Sovereignty must not be considered in isolation from economic viability:

“Sovereignty is only sustainable if it is economically viable.”

What a Successful Europe of the Future Could Look Like

For the next five to ten years, Khubbueva sees a Europe that:

• strengthens the internal market

• completes the capital market union

• actively builds global partnerships

• utilizes its industrial base

• socially accompanies transformation processes

Germany plays a key role in this – through its industry, research, and innovative strength.

Why Optimism is Still Appropriate

In conclusion, Khubbueva emphasizes what makes her most positively optimistic:

• ambitious startups

• innovation-driven companies

• common positions from both employer and employee sides

• industrial best practices that implement AI sensibly

• lived cooperation between works councils, employees, and technical teams

“The vision of a new future is actually shared by everyone.”