You recently wrote the report Applying AI in Key European Industries for Sitra. What are the main barriers to the adoption of AI in Europe?
Unfortunately, our starting point is not optimal. Western European companies lag behind US companies in AI and IT spending, and the gaps range from 45% to 70% across all sectors. In tech development, Europe holds a fair market share in AI semiconductors and service development but is not self-sufficient in critical AI raw materials, cloud infrastructure, and supercomputing capabilities.
Further, since 2022, over 90% of funding for large language model development has taken place outside Europe. European companies contributed only 25 of the 101 AI models recognized as notable, compared to 61 models from US companies. Therefore, if we don’t change course, we risk falling behind in the global competition.
What role do AI agents play in business and society, and what opportunities do they offer?
There is significant buzz around AI agents now. Some people have even begun referring to a “billion agent future” where a massive number of these agents work in the background, supporting us without our direct awareness. Unlike earlier AI models that typically assisted humans by merely suggesting actions, agents may be able to complete entire tasks rather autonomously.
AI agents are transforming companies by enabling scalability, automation and efficiency, with the potential to replace entire teams or empower lean organizations to scale rapidly in the future. They integrate into workflows, autonomously handling specialized tasks and facilitating a shift toward interconnected ecosystems like the agentic web – in a sense, a more developed version of the internet we now know.
At the societal level, AI agents enable new types of collaboration between humans and AI, including acting on behalf of their users in many tasks and potentially reducing human involvement in certain processes.
The report mentions that Europe lags behind the US and China in AI investments. What actions do you recommend to close this gap?
The European Union has very recently taken a decisive step in AI investments with the launch of the 200-billion-euro InvestAI initiative, announced in February 2025. This funding program aims to supercharge Europe’s AI capabilities by mobilizing capital for AI gigafactories, large-scale research, and development hubs designed to advance cutting-edge AI models.
The model is a public-private partnership similar to CERN – the European Organization for Nuclear Research (Conseil Européen pour la Recherche Nucléaire in French). CERN is the world's largest particle physics laboratory, which operates through multi-national funding and collaboration. This could potentially lead to the creation of new 'AI CERNs' for us.
InvestAI will provide scientists, startups, and companies of all sizes with the resources needed to develop AI models. Consequently, it seems that significant financing has been secured, and the next critical step is to allocate it quickly and efficiently to areas such as AI infrastructure and competence development so that it translates to tangible progress in AI development.
What are the two key areas of applied AI where European companies have the best chance to compete globally, and why?
It seems that relatively soon, nearly all companies will be using standard, commonly available AI systems. They will no longer be a source of competitive advantage even if they deliver significant productivity gains. Therefore, a different perspective is called for. While the global AI race has been dominated by the “bigger is better” paradigm, pushing companies to develop ever-larger models, the EU should adopt a more balanced approach.
Instead of competing purely on scale, Europe can carve out a competitive advantage by focusing on smaller, domain-specific AI models: vertical AI and AI agent applications in specialized use cases.
Vertical AI refers to artificial intelligence applications tailored to solve specific industry challenges and streamline processes.
There is a global market opportunity for vertical models and agents in:
- digital solutions for the mass consumer market (companies like Uber, Spotify, and Airbnb)
- less obvious and hard-to-predict mass consumer solutions (such as future AI-driven travel booking or shopping services)
- highly specialized, AI-powered, and AI-native solutions for business-to-business transactions (such as AI-driven sales, marketing, and product development services)
Of these, the last two seem the most promising for European companies.
How can Europe ensure its strategic autonomy in AI, especially concerning critical technologies like cloud infrastructure and semiconductors?
This could happen in many ways. For instance, Europe could support its private companies in these fields to help them stay competitive and grow both within and beyond the European market. These companies currently have relatively low brand recognition and a small market share. Public investments should also play a strategic role in developing critical AI technologies, making sure Europe doesn't fall behind in key areas.
On top of that, the European public sector should lead by example by being the first to adopt innovative European AI technologies and services. That said, there seems to be no quick or easy fix to achieving strategic autonomy. This transformation will probably take time.
What are your report’s main recommendations for strengthening Europe’s competitiveness and innovation in the field of AI?
The report presents six recommendations. I am proposing deeper European single market harmonization, including the removal of the remaining legal and technical barriers to data sharing and reuse. We need to ensure affordable and ecologically sustainable computing resources for the creation of AI applications, since AI’s energy consumption is still a considerable challenge that needs to be addressed.
In addition, we need to make large language models available in all official EU languages. European data spaces and ecosystems must become reliable sources of high-quality data for AI models, including data from public-sector sources. We need training programs that upskill the European workforce in AI, and we also have to do a better job of attracting and retaining top global AI talent.
Last but not least, as I mentioned earlier, we need to turn European AI investments into real action, because funding plays a critical role in implementing all of the recommendations above.
More information:
