The State of AI in Germany

Zane Davis

Halle Foundation/AGI Intern

Zane Davis is a summer 2025 Halle Foundation Intern at the American-German Institute. His primary research interests include artificial intelligence policy, labor economics, and global trade.

Prior to joining AGI, Zane completed his bachelor's degree in economics from the University of North Carolina at Chapel Hill, where he also earned minors in German and philosophy. He developed an interest in artificial intelligence and global trade while working for a Munich-based software startup in 2022. In 2023, he relocated to Germany, where he played American football for the Ingolstadt Dukes of the German Football League. Since then, he has worked for organizations spanning the United States, Korea, and Thailand, strengthening his background in public policy and international affairs.

In the fall of 2025, he will return to Chapel Hill to begin the Master of Public Administration program at the UNC School of Government.

Germany was an early adopter of AI and among the first nations to create a national AI strategy in 2018. At the time, Europe was the global leader in artificial intelligence research publications, ahead of both the United States and China. In industry and research, Germany was a leader in Europe, pledging more than double the public investment into AI as the United Kingdom or France and serving as the home of more than 100 AI startups. The 2018 national strategy enacted under then-Chancellor Angela Merkel offered a framework for securing Germany’s status as the base for trustworthy and sovereign AI.

Despite being an early mover on initiatives to regulate AI and a leader in academic research, Germany has failed to convert this early enthusiasm into sustainable growth in private sector deployment. Today, Germany has no frontier model, weak public investment, poor data and energy infrastructure, restrictive deployment regulations, and lagging capital expenditures. Compared to other European countries like Denmark, Finland, and the Netherlands, Germany’s overall AI adoption rate in businesses is falling behind.

Still, Germany has retained some of its early advantages. Germany’s research environment remains a strength, and initiatives connecting universities and research institutions with private sector actors have led to the creation of dozens of so-called “innovation clusters“ across the country, comprising more than 300 research institutions and over 100 “transfer centers” meant to close the gap between theory and practice. AI integration in healthcare and industrial manufacturing has been particularly strong, including government initiatives like the Health Data Lab at the Federal Institute for Drugs and Medical Devices and private sector developments led by Siemens, Bosch, and SAP.

Quietly, Germany is attracting net-inflows of AI-skilled labor at a rate even higher than the United States. Germany has also been steadily improving its status as a destination for AI founders, even surpassing powerhouses like China and France in global rankings. To accelerate these developments, Chancellor Friedrich Merz has announced his intention to limit unnecessary regulation, lower barriers to entry for AI firms, and commit to a lenient interpretation of the European Union’s AI Act.

Leaders at the sub-national level are chipping in, too. In total, there are more than forty national and regional state programs across Germany built to aid German businesses in implementing artificial intelligence. These efforts are all part of a broader strategy laid out by the Federal Ministry of Research, Technology, and Space to reclaim Germany’s “technological sovereignty” in response to growing dependencies on the United States, Russia, and China.

Past Ambitions

In mid-2018, more than four years before the release of ChatGPT, Chancellor Merkel was already advocating for a unified European strategy on artificial intelligence in line with her vision for German AI leadership. Her government’s emphasis on digital sovereignty, data privacy, and AI safety later became central components of the EU’s AI Act. By June of 2020, the governing coalition of Christian Democrats (CDU/CSU) and Social Democrats (SPD) had raised the federal government’s commitment to invest in AI initiatives from 3 to 5 billion euros and promised to redouble Germany’s efforts to remove the “market barriers” preventing Germany’s AI research capacity from translating into innovation in the private sector.

Despite these ambitions, the OECD’s 2023 review of Germany’s progress on AI objectives found that, “compared to other European countries, Germany lags behind.” Despite leading the EU in private investment into AI startups, the German government struggled to meet its own adjusted spending targets. Meanwhile, countries like France raced ahead, announcing a 2.2 billion euro investment package to foster new models and compute capacity.

Following the decades-long disappointment of the German digital transition, Germany risks falling behind on yet another global technological revolution.

In 2024, the European Parliament passed the AI Act—the world’s first comprehensive legal framework for regulating artificial intelligence. While Berlin had long advocated for  risk-based frameworks to regulate AI, German industry leaders began voicing concerns that the AI Act’s bureaucratic complexity and regulatory uncertainty would stifle innovation. Germany’s new governing coalition of the CDU/CSU and SPD has emphasized that the AI Act should be enforced in an “innovation-friendly” manner.

Present Challenges

In a broad sense, Germany is falling behind on AI. The highest-performing German large language models do not even register in the top 250 on most AI leaderboards, and benchmarking statistics suggest they are roughly comparable to models trained five or six years ago in the United States like OpenAI’s davinci-003.

Without a domestic frontier model, German businesses and private citizens primarily rely on American-built models like OpenAI’s GPT, Google’s Gemini, and Meta AI’s Llama. Even with advancements in open-source modeling like DeepSeek’s R1 (China) and Mistral’s Magistral (France), American models dominate most enterprise applications. The difficulty of regulating AI capabilities at the deployment stage makes this dependency on foreign models a legitimate concern for policymakers in both Brussels and Berlin.

Data

Other concerns include the lack of a dedicated AI computing infrastructure, which makes training and servicing models in Germany both slower and more expensive than in the United States. This trend is likely to continue, with 2024 private sector investment in AI compute capacity in Germany totaling only $54 million according to OECD estimates (a figure dwarfed by the nearly $2 billion invested in Canada). Germany’s total compute investment from 2020 to 2025 does not even reach half the investment volume of South Korea or Israel.

In addition, the majority of AI-specialized data centers in Germany are already hosted by American companies, and NVIDIA’s monopoly in the market for GPUs means that any future data centers in Germany will rely heavily on American-made chips. Ambitious, German-led projects like Gaia-X have attempted to build a federated European data infrastructure, but have struggled to gain traction against the market power of established U.S. providers.

Considering Germany’s pre-existing dependencies on foreign oil, critical minerals, and digital services, this reliance on American and Chinese firms is especially concerning. In 2023, the European Commission discovered that Europe relied on foreign countries for “over 80 percent of digital products, as well as for services, infrastructures, and intellectual property.” This growing pattern of technological dependency has become a source of considerable strategic anxiety in Berlin.

Energy

Even if Germany were able to catapult itself to the forefront of data center and compute infrastructure investment, the energy demands of next-generation AI data centers would still impose a significant burden on Germany’s energy grid. Following the German government’s transition away from Russian oil and gas imports, Germany’s energy-industrial base is straining. Supply constraints have resulted in energy prices more than 30 percent higher than the EU average. Securing multiple gigawatts of additional capacity on the rapid timeline demanded by the AI industry would be incredibly challenging. To put the energy demand of modern AI data centers in perspective, building a single data center on the scale of Mistral’s planned Paris cluster would require an additional 1.4 gigawatts of energy from Germany’s grid annually—roughly equivalent to the power consumed by one million homes.

Research and Industry

Despite these challenges in the private sector, Germany still places third in the world (only trailing the United States and China) in highly-cited AI research publications thanks to its strong university research ecosystem and leading institutions like the German Research Center for Artificial Intelligence (DFKI) and Cyber Valley. Where Germany lags in large-scale consumer AI, it has found success in some of the specialized applications of industrial AI, including predictive maintenance and supply chain optimization. In 2024, Germany led Europe in AI patent filings.

Future Policy Benchmarks

Germany is home to more than 600 AI startups, primarily concentrated in healthcare, logistics, and industrial robotics. Among industry leaders across sectors, concerns about overregulation continue to dominate discussions about the future of AI in Europe. In response to these concerns, Chancellor Merz intends to eliminate bureaucratic hurdles to private investment and increase public investment in essential energy infrastructure in the future. The German government has also pledged to spend at least 3.5 percent of its GDP annually over the next five years to bolster critical technologies, including investments in AI, quantum computing, and robotics.

Overall, the essential challenges faced by German leaders in AI are stimulating private investment, establishing energy security, and attracting competitive compute infrastructure. To make Germany a more attractive destination for capital investment in AI infrastructure including large-scale data centers and gigafactories, German leaders will need to create more attractive public-private partnership models for infrastructure development, streamline bureaucratized approval processes, and work to ensure more competitive and stable energy prices.