At Rise of AI 2024 in Berlin, Tina Klüwer, five months into her role at the German Federal Ministry of Research and Education (BMBF), explains how Germany turns AI research into industry applications. She traces the government's AI strategy from its 2018 launch, one of the first national AI strategies worldwide, through its 2020 update into six pillars including talents, research, transfer and applications, world of work, frameworks and standards, and society, backed by a total budget of 3.5 billion euros. She shows that Germany has a strong research base, including over 90 AI research institutions, the 1988-founded DFKI, and the 1997 LSTM breakthrough by Hochreiter and Schmidhuber, but ranks only 18th of 27 EU states at turning that research into applications. She presents the Ministry's Action Plan AI, published November 2023, with concrete transfer measures: SME funding, regional AI service centers, expanded GPU computing access, and talent programs including new AI professorships and the AI Campus platform.
Key takeaways
- Germany's 2020 AI strategy update consolidated 12 working areas into six pillars backed by 3.5 billion euros total funding.
- Germany ranks only 18th of 27 EU member states at transferring AI research results into practical applications.
- An Earlybird VC study found that only an estimated 25% of German AI research results are translated into applications.
- Only 15% of German companies currently report using AI, while 40% of German AI startups are academic spin-offs.
- The Action Plan AI, published November 2023, funds SME programs, regional AI service centers, and expanded GPU access including a Julich supercomputer reaching about 25,000 GPUs by early 2025.
Detailed notes
Germany's AI strategy from 2018 to today
Klüwer recounts that Germany published one of the world's first national AI strategies in 2018, organized around 12 working areas covering research, transfer and business, skills training, standards and labor research. A 2020 update consolidated this into six pillars: talents, research, transfer and applications, world of work, frameworks and standards, and society. The government allocated a total of 3.5 billion euros to these pillars, with roughly 2 billion euros going to transfer and application, and about a billion to research, the pillar she is now responsible for.
A strong research base with a transfer gap
Klüwer describes a long German AI research history, from early language-research projects to the founding of the German Research Center for Artificial Intelligence (DFKI) in 1988, and the 1997 long short-term memory (LSTM) result by Hochreiter and Schmidhuber, now a foundation of neural networks including generative AI. Germany now has more than 90 AI research institutions and university departments, plus six newer AI competence centers. Yet on 2023 publication counts Germany ranks fifth behind China, the US, India and the UK, and for newly funded AI companies and AI patents it ranks ninth. Overall, Germany ranks only 18th of 27 EU member states at transferring research results into practice; an Earlybird VC study found only an estimated 25% of research results are translated into applications.
Current state of AI adoption and startups
Klüwer states that only 15% of German companies currently report using AI, calling the number low and pointing to the potential if adoption reached 50%. Germany has roughly 500 AI startups, a growing number, and 40% of them are academic spin-offs compared with 2.4% for startups generally, showing a strong link between research and founding activity. On AI patents Germany ranks mid-tier, behind the US, China, South Korea and Japan.
The Action Plan AI and transfer measures
The Ministry of Research published the Action Plan AI in November 2023 as an implementation of the existing AI strategy, organized into 11 fields of action with transfer as a central goal. Concrete measures include a new SME funding program, a regional competence center for labor research in AI, and four AI service centers in Berlin-Brandenburg, Lower Saxony, Hessen and Ostwestfalen offering SMEs advice and support. The government also offers startups and SMEs low-threshold access to about 1,000 GPUs, plus a link to the Julich Supercomputing Center, which is being expanded to roughly 25,000 GPUs by early 2025 as one of the largest GPU supercomputers in Europe, alongside 10 to 20 million euros of additional SME funding.
Building AI talent
Klüwer describes a competency offensive that added AI professorships across German universities, calls for junior scientists with a focus on women, and the AI Campus learning platform for the broader public. She closes by noting that Germany now ranks third worldwide in AI skill penetration rate, behind only India and the US.
Quotes
Germany is one of the largest producers of scientific results worldwide, but we only rank 18 out of 27 EU member states for transferring these results into practice.
Tina Klüwer
We now do have in Germany more than 90 AI research institutions and university departments all distributed around Germany.
Tina Klüwer
Currently, we only have 15% of all German companies saying that they use AI.
Tina Klüwer
Questions answered in this talk
What percentage of German companies use AI?
According to Tina Klüwer at Rise of AI 2024, only 15% of German companies currently report using AI.
What is Germany's Action Plan AI?
According to Tina Klüwer at Rise of AI 2024, the Action Plan AI is a Ministry of Research plan published in November 2023 that implements Germany's existing AI strategy through 11 fields of action, with transfer of research into industry as a central goal.
How much of German AI research is actually turned into applications?
According to Tina Klüwer at Rise of AI 2024, citing an Earlybird VC study, only an estimated 25% of German AI research results are translated into applications, and Germany ranks 18th of 27 EU states on research transfer.