Joint Feedback from the German AI Association and the
European AI Forum on:
Apply AI Strategy:
Strengthening the AI Continent
Today’s breakthroughs in AI promise a transformation as sweeping as the rise of the internet. The decisive question is who will control the technology, data, and resources required for deployment at scale. At this juncture, the European Union holds a critical stake in shaping AI’s global course. As international competition intensifies, the EU’s capacity to secure economic and strategic autonomy will depend on its ability to establish digital sovereignty and enhance its competitiveness. Only those who have mastered the technology will be able to shape its use according to their own ideas, thereby positively influencing economic and social developments.
To achieve and sustain a leading position in the global AI landscape, the European Union and its member states must be able to:
1. Develop and market state-of-the-art, globally competitive AI models/systems; 2. Provide an attractive ecosystem for AI talent and founders;
3. Increase the implementation of AI solutions across the entire economy;
4. Decrease its dependency on non-European AI solutions and infrastructure.
We therefore appreciate the European Commission's efforts to address these issues through the AI Continent Action Plan. Although we have previously raised concerns about the plan as a whole, we very much welcome the Commission's intention to develop an 'AI Strategy', as we believe this to be crucial to establishing the EU as an AI continent. We welcome the opportunity to provide feedback and submit a variety of policy recommendations.
The following feedback-paper is structured into two parts. The first part is a review of the current stage of AI in the European Union, focusing on the development and implementation of AI in the EU. The second part provides our recommendations to the European Commission for the Apply AI Strategy.
Page 1 of 13
1. Artificial Intelligence in the European Union
The following chapter will map the current state of AI development and deployment, as well as identify key barriers to scaling up and highlight structural gaps in innovation, talent, and infrastructure.
1.1.Development of AI in the European Union
The EU holds a nuanced position in the global AI landscape. Although it is well-positioned at the forefront of the global research and innovation ecosystem, private-sector investment and innovation in the EU lag behind those of leading AI “powerhouses” such as the United States and China. For example, the EU’s share of global ICT market revenues has halved in the last 10 years, only reaching 11 percent in 2022, and it currently relies on non-EU countries for 80 percent of digital services and products, as well as infrastructure and intellectual property.1 Instead, recent years have been a testament to the EU’s commitment to regulation. While competitors such as the United States have focused on fostering their innovation and advancing this key technology, the EU has primarily emerged as the leading authority on AI regulation.
1.1.1. Investment Disparities and Strategic Gaps
When it comes to private investment in AI, US private investors collectively invested around $67.22 billion in 2023. In contrast, private investment in AI in the EU and UK combined only amounted to $11 billion. Germany was the highest-ranking EU member state, investing $1.91 billion.2
Furthermore, although private investment in the US has increased significantly since 2022, the EU has experienced a second consecutive year of declining AI investment. In terms of private investment in generative AI specifically, the EU only saw a modest increase to €0.74 billion, whereas private investment in generative AI in the US surged to $22.46 billion.3
1 European Commission. (2023). 2023 Report on the state of the Digital Decade: Shaping Europe’s Digital Future.
https://digital-strategy.ec.europa.eu/en/library/2023-report-state-digital-decade
2 Maslej et al. (2024). The AI Index 2024 Annual Report. AI Index Steering Committee, Institute for Human-Centered AI, Stanford University. https://aiindex.stanford.edu/wp-content/uploads/2024/04/HAI_AI-Index-Report-2024.pdf
3 Maslej et al. (2024). The AI Index 2024 Annual Report. AI Index Steering Committee, Institute for Human-Centered AI, Stanford University. https://aiindex.stanford.edu/wp-content/uploads/2024/04/HAI_AI-Index-Report-2024.pdf
Page 2 of 13
1.1.2. Research Output and Technological Specialisation
The EU combines a long record of academic excellence with a large reservoir of AI talent, and in recent years it has secured a prominent spot on the global stage. However, it still faces challenges such as insufficient funding and bureaucratic complexities, that hold back its full potential
Furthermore, translating this scientific prowess into lasting economic and technological leadership is fraught with challenges. Another obstacle to academic excellence in AI is the lack of secure, open data pools. Additionally, differing open data pool policies across Member States result in inefficiencies for researchers and hinder innovation.4 Moreover, in terms of research publications, China leads AI patents (2014–2023), far surpassing both the U.S. and the EU.
1.1.3. Strategic Infrastructure Development
The EU’s mounting reliance on non-European technology, especially in AI, raises serious concerns, nowhere more evident than in its dependence on external compute infrastructure. Currently, European cloud companies account for around five percent of the global market - meanwhile, US hyperscalers account for around 85 percent. Although Germany has improved its position in recent years, it still lags behind other EU member states and is comparably even further behind the US and China. This oftentime leaves European enterprises with no option but to choose non-European solutions, which further harms the EU’s digital sovereignty.5
The European Commission has put forward initiatives to address these shortcomings, such as AI factories and gigafactories. While these initiatives represent a long-overdue step towards reducing dependence on foreign cloud and AI infrastructure, enabling European actors to develop and train large-scale AI models locally, they do come with some constraints and structural deficiencies, particularly for private enterprises.6
4 Konrad-Adenauer-Stiftung e.V. (2020). Analysis of current global AI developments with a focus on Europe. https://www.kas. de/documents/252038/7995358/Analysis+of+current+global+AI+developments+with+a+focus+on+Europe.pdf/da00b869c c82-03e4-a45d-b74357fdd9c1?version=1.0&t=1606820629157
5 McKinsey Global Institute (2024). Time to place our bets: Europe’s AI opportunity.
https://www.mckinsey.com/capabilities/quantumblack/our-insights/time-to-place-our-bets-europes-ai-opportunity 6 We will shortly be publishing a position paper on AI infrastructure in the EU.
Page 3 of 13
1.1.4. AI Talent Dynamic
Although the number of AI and data scientists, as well as graduates in these fields, has increased in the EU in recent years, companies in the EU are still struggling to fill vacancies in the ICT sector. There is a particular shortage of talent in the AI sector, and it is estimated that there will be a deficit of 12 million skilled workers by 2030. The direct consequence: EU companies are hindered in their investment activities.7
Currently, there is a visible pattern of innovation and intellectual property moving to markets outside the EU. This is particularly evident among AI-skilled graduates who leave the EU to work in non-EU markets. Germany, for example, has seen over half of its doctoral students in AI-related fields leave the country after graduation, primarily towards the US, the UK and Switzerland.8 The EU's "brain drain" therefore remains acute, posing a significant challenge to maintaining economic value and influence in the global development and application of AI technologies.
1.1.5. Transfer of Research and Spin-Offs
AI start-ups and spin-offs are among the most promising ways of translating research results into practical applications. Start-ups are, by definition, focused on growth and scaling their business models, so a spin-off will automatically scale up and maximise the impact of the underlying scientific achievements. Therefore, it is important to support the start-up environment to stimulate innovation and strengthen the transfer of research into spin-offs.
However, the transfer of research into practical applications is often hindered by structural barriers such as differing IP policies across EU member states, bureaucratic inefficiencies and a lack of funding opportunities.9 Thus, both the EU and its member states are looking to realise the untapped potential that could significantly improve their standing as global AI players.
7 European Commission. (2023). 2023 Report on the state of the Digital Decade: Shaping Europe’s Digital Future.
https://digital-strategy.ec.europa.eu/en/library/2023-report-state-digital-decade
8 Stiftung Neue Verantwortung. (2023). AI Talent Flows in Germany: Empirical study of the career paths of AI doctoral students at German universities. https://www.stiftung-nv.de/sites/default/files/230112_snv_talentflowanalyse_eng_final.pdf 9 German AI Association (2023). KI-Startups und Wissenschaft.
https://ki-verband.de/wp-content/uploads/2023/06/KI-Startups-Wissenschaft.pdf
Page 4 of 13
1.2.Deployment of AI in the European Union
EU companies average at approximately 13.5% in terms of usage and deployment of AI solutions. Around 20 percent of enterprises in Germany have integrated AI solutions into their work, placing the member state above the EU average.10Significant differences in the adoption of AI can be observed among enterprises of different sizes. While large businesses with 250 or more employees have comparatively high adoption rates, smaller businesses with fewer than 250 employees are deploying significantly fewer AI solutions.11
Although Europe’s AI adoption is growing continuously, they still lag far behind the US, where at least 70 percent of businesses use AI for at least one business function. Other countries with high adoption rates include India (59 per cent), China (50 per cent) and the United Arab Emirates (58 per cent).12
1.2.1. Adoption Rates and Market Penetration in Different Sectors
However, these aggregate figures mask significant variations in national and sectoral adoption patterns. The following analysis therefore highlights the use of AI in different sectors. Although the main focus is on the German market, comparisons are also made with other EU member states.13
High-Adoption Sectors:
â—Ź The ICT sector is leading the way in AI adoption across Europe, with rates ranging from 21.0% in Italy to 48.2% in Finland. Germany's adoption rate in this sector is 33.1%.
â—Ź Consulting, legal and technical services consistently demonstrate high adoption rates across EU countries. Here, Germany leads the way with an adoption rate of 26.3% in advisory services (legal and tax consulting, engineering firms and R&D services), followed by Finland with 31.5%.
â—Ź Banking and financial services also demonstrate strong AI adoption, particularly in Germany (34% for banks) and other developed European markets.
10 Eurostat (2025). Usage of AI technologies increasing in EU enterprises.
https://ec.europa.eu/eurostat/web/products-eurostat-news/w/ddn-20250123-3 11 Euronews (2025). AI on the rise among European businesses: How are they using it?
https://www.euronews.com/my-europe/2025/01/31/ai-on-the-rise-among-european-businesses-how-are-they-using-it 12 Forbes (2024). 22 Top AI Statistics And Trends. https://www.forbes.com/advisor/business/ai-statistics/#:~:text=AI%20is%20expected%20to%20see,technologies%20in%20the%20coming%20years.
13 Sources for the following chapter include: Bundesministerium fĂĽr Wirtschaft und Klimaschutz (2024). KI-Einsatz in Unternehmen in Deutschland - Strategische Ausrichtung und internationale Position. https://www.de.digital/DIGITAL/Redaktion/DE/Digitalisierungsindex/Publikationen/publikation-ki-einsatz-2024.pdf?__blob=publicationFile&v=2; Euronews (2025). AI on the rise among European businesses: How are they using it?. https://www.euronews.com/my-europe/2025/01/31/ai-on-the-rise-among-european-businesses-how-are-they-using-it
Page 5 of 13
Medium-Adoption Sectors:
â—Ź Traditional manufacturing shows a varied adoption pattern. The chemical, plastics, and construction materials industries have an average adoption rate of 8.8% across all EU member states. The electrical, machinery and automotive industries are similar, with an EU average of 8.7%, while the metal industry has a lower average of 5.7%.
â—Ź In the energy and utilities sector, the EU average adoption rate is 8.8%, with the frontrunner being Denmark at 27.7%.
Low-Adoption Sectors:
â—Ź The construction sector shows consistently low adoption rates across Europe with an average adoption rate of 3.2%.
â—Ź Similarly, the food industry remains untapped potential with only 5.0% EU average adoption.
1.2.2. Public Sector Implementation
European public administrations should be flagship users of AI, raising service quality for citizens and streamlining internal processes. Yet adoption remains low, and the public sector rarely acts as an anchor customer - depriving domestic AI firms of a vital early market and eroding Europe’s strategic edge.14 In the US, for instance, public sector procurement provides significant market opportunities for American AI start-ups, whereas European public administration has yet to achieve comparable levels of AI adoption.15
1.2.3. Technology Deployment Patterns
Speech recognition and generation are currently the most widely deployed AI technology across the EU.16 Germany leads this trend at 5.0% of all enterprises compared to the 2.5% EU average. Workflow automation and decision-making support technologies are adopted by approximately one-third of AI-using companies, reflecting a focus on operational efficiency over innovative product development. Text analysis and mining technologies are
14 Capgemini Research Institute (2025). Data foundations for government - From AI ambition to action.
https://www.sogeti.com/wp-content/uploads/sites/3/2025/05/Capgemini-Research-Institute-report_Data-foundations-for-government_From-AI-ambition-to-execution-3.pdf
15 German AI Association (2024). FĂĽr ein starkes KI-Deutschland - Impulspapier zur Bundestagswahl 2025.
https://ki-verband.de/wp-content/uploads/2024/12/Impulspapier_Bundestagswahl2025_KI-Bundesverband_2024.12.pdf 16 Eurostat (2025). Usage of AI technologies increasing in EU enterprises.
https://ec.europa.eu/eurostat/web/products-eurostat-news/w/ddn-20250123-3
Page 6 of 13
gaining traction, particularly in knowledge-intensive industries where regulatory compliance is crucial.17
When examining application areas, the majority of AI deployment occurs within products and services themselves, with 68% of German AI-using companies integrating AI directly into their offerings. This pattern remains consistent across leading European markets. Internal processes represent the second most common application area, with 54% of companies deploying AI for production, logistics, and service delivery operations. Sales, marketing, and customer contact applications account for approximately 27% of AI implementations among companies using this technology.18
1.2.4. Market Integration and Adoption Barriers
European AI companies face substantial challenges in achieving market penetration within their domestic markets. Despite overwhelming recognition of AI's value proposition, the vast majority of European companies have not implemented AI solutions, creating a paradoxical situation where domestic demand remains largely unrealized. This adoption lag threatens to drive AI market development toward international competitors, potentially undermining European AI and digital sovereignty objectives.
Skills and Human Capital Constraints:
One of the most significant barriers to the adoption of AI across European enterprises is the shortage of skilled personnel. Around 80% of firms cite a lack of skills as a major obstacle, encompassing limitations within their own workforces and constraints within the wider labour market.19 This skills gap manifests at multiple levels. For example, European firms are finding it difficult to recruit AI talents who can adapt algorithms to specific operational needs. Beyond core AI expertise, enterprises lack personnel with the hybrid skills needed to integrate AI systems into existing business processes. These include data engineers, business analysts with AI literacy, and managers who understand technology capabilities and business requirements.
17 Startbase (2024). 20% of German companies will be using AI technologies in 2024.
https://www.startbase.com/news/20-der-deutschen-unternehmen-nutzen-2024-ki-technologien/; Eurostat (2025). Usage of AI technologies increasing in EU enterprises.
https://ec.europa.eu/eurostat/web/products-eurostat-news/w/ddn-20250123-3 18 ifo Institute (2024). More Companies in Germany Using Artificial Intelligence.
https://www.ifo.de/en/facts/2024-07-18/more-companies-germany-using-artificial-intelligence 19 Hoffmann and Nurski (2021). What is holding back artificial intelligence adoption in Europe?
https://www.bruegel.org/sites/default/files/private/wp_attachments/PC-24-261121.pdf
Page 7 of 13
Financial and Resource Constraints:
Financial barriers present another major hurdle, particularly for SMEs. The costs associated with an initial AI adoption project extend beyond the acquisition of technology. Unlike large enterprises, which can spread costs across multiple use cases, SMEs often find it difficult to warrant the initial investment. Secondly, implementing AI necessitates significant changes to business processes, employee training, and organisational structures. Lastly, AI systems require robust IT infrastructure, including data storage, processing power, and networking capabilities.
Data and Infrastructure Limitations:
AI systems require extensive data collection, storage and processing capabilities, which are based on foundational digital infrastructure. Despite relatively high overall digitisation rates, significant gaps remain in business process digitisation. For example, only 33% of European companies use customer relationship management systems, 36% use enterprise resource planning software, and just 12% perform big data analysis. These foundational systems are prerequisites for effective AI implementation. Policy often spotlights access to external datasets, but the decisive hurdle sits inside the firm: reliable, well-structured internal data. Lacking rich historical records, companies cannot train or adapt AI models, and few maintain the data-governance practices required to generate them. Even where data exists, it often lacks the required structure, completeness or quality for AI systems. Legacy systems create data silos and inconsistent formats that must be resolved before AI implementation can proceed.
Regulatory and Legal Uncertainty:
European enterprises face significant barriers to AI adoption, primarily due to regulatory uncertainty and organisational readiness issues. For instance, the phased implementation and current delays of delegated acts, guidelines and code of practices, is leading to EU firms postponing AI projects until anticipation of clearer guidance.20 Furthermore, this regulatory environment forces European companies to allocate significantly more of their AI budgets to legal compliance, compared to their US counterparts, thereby diverting funds away from research and development.21 In addition to that, the demanding European data protection framework creates unique implementation challenges. Furthermore, different industries face varying regulatory requirements that may conflict with compliance requirements from the AI Act.
20 Bain & Company (2024). Survey: Generative AI’s Uptake Is Unprecedented Despite Roadblocks.
https://www.bain.com/insights/survey-generative-ai-uptake-is-unprecedented-despite-roadblocks/
21 Kilian, Jaeck, Ebel (2025). European AI Standards – Technical Standardization and Implementation Challenges under the EU AI Act.
https://ki-verband.de/wp-content/uploads/2025/03/Study_European-AI-Standards_FINAL_20250325.pdf
Page 8 of 13
2. Strengthening the AI Continent: Recommendations
As the previous analysis demonstrates, the EU is facing critical challenges in maintaining its competitiveness in the development and deployment of AI. To address these challenges, we advise integrating the following policy recommendations into the Apply AI Strategy to strengthen Europe's position in the global AI landscape and boost the deployment and usage of European AI solutions by enterprises and the public sector.22
1. Establish a dual AI implementation voucher system
We recommend the introduction of an AI Voucher23 which is designed to encourage SMEs to implement AI solutions by reducing the financial risks associated with such projects. It provides financial support or subsidies for the development of AI applications in cooperation with European AI companies. By encouraging a wider range of industries to invest in AI solutions developed in the EU, it encourages direct investment in AI and highlights the role of governments in creating an environment conducive to innovation.
This system would launch AI Implementation Vouchers, covering up to EUR 50,000 for SMEs commissioning AI prototypes and pilots from European AI start-ups. At the same time, it would deploy AI Compute Vouchers, providing up to EUR 50,000 in credits for model training on European AI infrastructure, such as AI Factories or, potentially, AI Gigafactories in the future.
This approach addresses the critical financial barrier that prevents SMEs from experimenting with AI, while ensuring that data and development remain within European infrastructure.
2. Simplify AI regulatory requirements and extend compliance deadlines
In light of the slow development of harmonised standards for high-risk AI systems, as well as general delays in implementing the AI Act, we strongly recommend extending the existing deadlines and simplifying the regulatory framework. Companies report that they need at least twelve months to comply with these standards.24 The delay in developing
22 Please note that a detailed position paper, including concrete policy recommendations on strengthening European AI infrastructure, will be published soon.
23 The original concept of an AI Voucher for Germany can be found here:
https://ki-verband.de/wp-content/uploads/2022/03/KI-Voucher_Vorschlag-KI-Bundesverband.pdf
24 Kilian, Jaeck, Ebel (2025). European AI Standards – Technical Standardization and Implementation Challenges under the EU AI Act.
https://ki-verband.de/wp-content/uploads/2025/03/Study_European-AI-Standards_FINAL_20250325.pdf
Page 9 of 13
these standards eliminates the most accessible compliance pathway, forcing companies to rely on expensive expert opinions. This creates an uneven playing field that disadvantages start-ups and SMEs, thereby further slowing down the deployment of AI use cases in the EU. To address these critical bottlenecks, we recommend extending the implementation deadlines. This timeline extension should provide realistic compliance periods that restore standards-based pathways as viable options for all organisations.
3. Develop an AI compliance cost reduction tool
Alongside the simplification of existing AI-related regulations, we recommend establishing an AI Compliance Cost Reduction Programme to minimise the regulatory burden on SMEs deploying AI solutions. This programme should provide automated AI Act compliance assessment tools to enable companies to efficiently evaluate their AI systems against regulatory requirements, reducing time and financial resources necessary. We recommend that the European Commission, through the AI Office, facilitates the creation of shared compliance services accessible to SMEs via industry associations. This would allow smaller companies to pool resources and expertise, rather than bearing the full cost of regulatory compliance infrastructure individually.
Furthermore, a dedicated 'AI Legal Tech' voucher programme should be launched to support the development of compliance software solutions, providing financial incentives for companies to create innovative tools that automate and simplify regulatory compliance processes. This comprehensive approach will prevent European companies from allocating excessive resources to regulatory compliance activities at the expense of research and development, ensuring they remain competitive with international competitors who may face less stringent regulatory environments.
4. Implement a strategic public procurement to boost European AI
To boost European AI innovation and significantly increase adoption, we recommend a strategic procurement policy that increases public-sector investment in AI and digital technologies. This policy should prioritise European suppliers to reduce dependency on non-EU AI solutions, thereby strengthening the EU's digital sovereignty. This should be achieved through a hybrid implementation strategy, whereby short-term contracts utilise existing market-leading technologies from global providers to address immediate digitalisation needs. At the same time, long-term procurement frameworks should be established to specifically support European companies in developing innovative AI solutions and foundational digital services.
Page 10 of 13
At the EU and member state levels, public institutions should actively function as anchor customers, providing a stable demand and revenue stream that will enable European AI companies to expand their operations and compete effectively in global markets. This strategic procurement approach must be coupled with increased overall budget allocation for AI and digital transformation projects at all levels of government, moving away from the current practice of centralising procurement through billion-euro framework agreements with a few non-European providers. Implementation should include regular assessment mechanisms to monitor progress towards reducing technological dependencies while maintaining the operational efficiency of digitalisation efforts in public administration.
5. Establish research-to-market initiatives to boost spin-offs
To better incentivise and facilitate the transformation of academic research into practical commercial applications, we recommend a comprehensive research-to-market transfer initiative to enhance spin-off creation and intellectual property commercialisation frameworks.
This initiative should involve embedding long-term support programmes directly within research institutions to provide scientists with dedicated resources, mentorship and financial incentives to help them evaluate the commercial potential of their research and develop viable business models around their innovations. Thus, universities should all offer graduate students and researchers entrepreneurship courses, equipping them with the essential skills needed to assess spin-off opportunities and navigate the commercialisation process effectively. Additionally, to address critical barriers in intellectual property transfer, IP negotiation frameworks must be improved and should be standardised across all member states. This would make the conditions for granting intellectual property rights more transparent and accessible for potential founders, while reducing the bureaucracy that currently impedes spin-off creation. Lastly, the initiative should provide dedicated funding for proof-of-concept projects and early-stage spin-offs, bridging the gap between academic research and market-ready products. Ultimately, this would strengthen Europe's innovation ecosystem and ensure that scientific breakthroughs translate into economic value and competitive advantage.
6. Create "Erasmus for AI" with Integrated Digital Skills Framework
In order to address the critical skills gap and brain drain, an 'Erasmus for AI' programme with an integrated digital skills framework should be created. This comprehensive programme for the mobility and development of AI talent would establish EU-wide
Page 11 of 13
exchange programmes for AI researchers and professionals, deploy digital skills wallets with standardised AI micro-credentials, implement fast-track visa systems for non-EU AI talent and launch diverse fellowship programmes. This initiative would counteract the significant brain drain and address the projected skills shortage in the EU by 2030.
7. Adopt a consequent “European AI-First” approach
Finally, we advocate that the Apply AI strategy should lead with deployment: scale European-developed AI tools that meet the concrete needs of European industry. Success must be tracked through hard metrics - the share of firms using EU-built AI, the number of live deployments in priority sectors, and measured gains in productivity and competitiveness. In today’s unsettled geopolitical climate, such focus is not optional; it is the bedrock of digital sovereignty and, by extension, Europe’s economic security and strategic autonomy.
Page 12 of 13
Contact:
Daniel AbbouAlessandro Blank President Head of Public Affairs European AI ForumGerman AI Association
Email:
About:
German AI Association
The German AI Association is Germany’s largest industry association for Artificial Intelligence (AI) and represents over 500 innovative SMEs, start-ups and entrepreneurs focusing on the development and application of AI. We support AI entrepreneurs by representing their interests in politics, business and the media. Our goal is an active, successful and sustainable AI ecosystem in Germany and Europe. After all, we can only compete globally if the brightest minds and visionaries decide to set up businesses, conduct research and teach in the European Union. Our members are committed to ensuring that AI technology is applied in accordance with European and democratic values and that Europe achieves digital sovereignty. To achieve this, the European Union must become an attractive place for entrepreneurs to do business, where their willingness to take risks is valued and their innovative spirit is met with the best conditions.
European AI Forum:
The European AI Forum (EAIF) connects the most innovative AI and Deep Tech companies across Europe with established industry and policymakers, representing over 2,000 AI entrepreneurs through our member organisations to form Europe's largest AI network. The members of the European AI Forum are committed to ensuring that AI technology is applied in accordance with European and democratic values, helping Europe achieve digital sovereignty. To achieve this, the European Union must become an attractive AI location for entrepreneurs, where risk-taking is valued and innovative spirit meets the best conditions.
Page 13 of 13