A policy analysis from July 2026
In November 2018, the German Federal Government launched its first National Artificial Intelligence Strategy, a joint effort by the education and research, economic affairs, and labour ministries that set out an ambition simple enough to fit on a banner: make Germany and Europe a leading centre for AI. Backed initially by 3 billion euros through 2025 and topped up with a further 2 billion euros from the 2020 Economic Stimulus and Future Package, the strategy promised research funding, new professorships, a network of competence centres, and a societal dialogue on AI ethics. Eight years, one pandemic, one change of government, and several trillion dollars of global AI investment later, Germany finds itself in a familiar and uncomfortable position: well-funded on paper, still catching up in practice.
That tension defines the country’s AI story heading into the second half of 2026. Chancellor Friedrich Merz’s coalition has poured political capital and fresh borrowing into a rebranded innovation push, AI adoption among German businesses has more than doubled in a single year, and a landmark merger has just created Europe’s most credible sovereign AI challenger. Yet the country’s own digital minister stood on stage this spring and told an audience of tech executives that Germany has “a lot of catching up to do” on AI, comparing German and American stock market capitalisation as a blunt measure of who is actually winning the race his country entered in 2018.
FROM AI STRATEGY TO AI ACTION
Germany’s original 2018 strategy rested on three goals: strengthen Germany’s competitiveness as a research and business location for AI, ensure responsible AI development that serves the public good, and integrate AI into society through broad dialogue. The 2020 update, following recommendations from the Bundestag’s AI Study Commission, refined the focus areas to research, talent, technology transfer, regulation, and societal engagement, while adding sustainability and pandemic response to the mix.
The Federal Ministry of Research, Technology and Space, now known by the acronym BMFTR under minister Dorothee Bär, issued an AI Action Plan in November 2023 that identified eleven priority fields of action, from funding AI centres of excellence to expanding computing capacity. That plan has since evolved again. The BMFTR now states it will invest more than 1.6 billion euros in AI over the current legislative period, funding six AI competence centres that anchor the national research ecosystem, alongside a growing network of AI service centres designed to give small and mid-sized firms access to compute, pretrained models such as LAION-5B and LeoLM, and data protection-compliant tools through initiatives like ChatAI.
The bigger structural shift came in mid-2025, when Merz’s government folded AI into a much larger vehicle: the High-Tech Agenda Germany, adopted in July 2025 and formally launched at a Berlin ceremony that October. The Agenda covers six key technologies, including AI, quantum computing, microelectronics, biotechnology, fusion energy, and climate-neutral mobility, and is funded through a rising allocation that climbs from 500 million euros in 2025 to 1 billion euros annually from 2026, totalling roughly 5.5 billion euros over the legislative period. Bär’s ministry alone commands a 2025 federal budget of 22.38 billion euros, of which 18.7 billion euros goes to grants and allocations and 3.9 billion euros to direct investment. Merz has framed the wider push, alongside private capital, as a 130 billion euro combined investment effort, with a broader industry-led initiative pledging over 800 billion euros by 2028.
THE INFRASTRUCTURE SPRINT
If there is one place where Germany’s AI ambitions have moved from paper to concrete, it is compute. In March 2026, Digital Minister Karsten Wildberger unveiled cabinet-approved targets to at least double Germany’s overall data centre capacity by 2030 relative to 2025 levels, and to quadruple capacity specifically earmarked for AI workloads. The policy paper behind those targets sets out 28 measures meant to make Germany a sovereign data hub, including taxing data centres where they physically operate rather than where their parent company is registered, streamlining planning approvals, and pushing new facilities toward renewable power. A flagship commitment is support for at least one commercial AI gigafactory on German soil, built through a public-private consortium under European leadership, competing in an EU-wide tender that has already drawn close to 100 applicants.
Deutsche Telekom has moved fastest on this front. Working with Nvidia and investment firm Brookfield, the carrier is building what it calls Europe’s first industrial AI cloud, based in Munich and expected to add roughly 50 percent to Germany’s existing AI computing power once operational in early 2026. The site includes around 20 petabytes of storage and 75 kilometres of fibre optic cable, with construction partly carried out by robots. Deutsche Telekom is separately eyeing North Rhine-Westphalia, in talks with utility RWE, for the larger gigafactory project tied to the EU tender.
Public research infrastructure has also expanded. The JUPITER exascale supercomputer at the Forschungszentrum Jülich, opened in September 2025 with Merz in attendance, now anchors an AI Factory that gives industry and researchers access to frontier-scale compute for AI research and industrial pilots. The federal government’s KIPITZ AI portal is being extended through the second half of 2026 to reach state and municipal authorities via the Deutschland-Stack, Germany’s planned sovereign cloud, with a KIPITZ 2.0 app-store model for approved AI tools, including agentic capabilities, expected to follow.
A SOVEREIGN AI CHAMPION EMERGES
The most consequential single development in Germany’s AI landscape this year did not come from Berlin at all, but from Heidelberg and Toronto. On April 24, 2026, Canada’s Cohere announced it would merge with Germany’s Aleph Alpha, creating a combined entity valued at roughly 20 billion dollars, with global headquarters in Toronto and a European centre of excellence in Berlin. The deal is anchored by a 600 million dollar commitment from German retail conglomerate Schwarz Group, which already backed Aleph Alpha and expects the combined company to run on STACKIT, its own sovereign cloud platform. SAP and Bosch, both existing Aleph Alpha investors and customers, sit on the deal’s periphery, with SAP chief executive Christian Klein publicly welcoming the tie-up as giving Europe “a sovereign frontier partner with delivery scale.”
The merger is explicitly framed around procurement politics as much as technology. Several German federal agencies and at least three federal states hold AI contracts with clauses requiring vendors to be controlled in Europe, and Cohere has committed in writing to keep European public-sector data, model weights, and inference traffic within the STACKIT sovereign perimeter. The deal, backed publicly by both the Canadian and German governments under a new Canada-Germany Sovereign Technology Alliance, gives Germany a credible answer to the sovereignty argument that has driven so much of the continent’s AI policy, even if the combined company’s ultimate ownership sits partly outside the EU.
ADOPTION FINALLY ACCELERATES, BUT UNEVENLY
For years, the most persistent criticism of Germany’s AI strategy was not about funding levels but about diffusion: research money flowed, competence centres opened, and yet German companies, especially the Mittelstand that forms the backbone of the economy, adopted AI more slowly than peers in the US, UK, and parts of Scandinavia. That picture has shifted meaningfully in 2026. According to the digital association Bitkom’s annual AI study, published in April 2026 from interviews with more than 600 companies, AI adoption among German businesses reached 41 percent this year, more than double the 17 percent recorded in 2025 and the fastest year-on-year jump since tracking began. Development bank KfW puts SME-specific adoption at 20 percent, up from just 4 percent in 2020, with roughly 780,000 small and mid-sized enterprises now using AI in at least one business process.
The adoption gap has not closed so much as changed shape. Large enterprises with compliance teams, dedicated data platforms, and vendor leverage continue to move faster than everyone else; sector adoption ranges from around 84 percent in advertising and market research down to roughly 31 percent in construction, based on 2025 business survey data. A Boston Consulting Group survey cited alongside the Bitkom figures found that 78 percent of companies now use generative AI in some form, and 52 percent plan to invest more than 50 million dollars in AI this year, suggesting the money is no longer the bottleneck it once was.
What has replaced funding as the constraint is execution. Industry researchers describe a widespread pattern of “pilot purgatory,” in which AI projects launch with enthusiasm but never reach production. One analysis citing Gartner projects that roughly 30 percent of generative AI pilots will be abandoned by the end of 2026 for reasons that have little to do with the technology itself: unclear success metrics, missing executive sponsorship, and no realistic plan to scale beyond the experiment. A parallel study from ISG puts the pilot-to-scale failure rate as high as 69 percent. One in three German companies surveyed by Bitkom reported that AI turned out to be more expensive than expected once token costs, GPU hosting, systems integration, and governance overhead were added to the bill. And even where AI is deployed, trust remains conditional: a spring 2026 YouGov survey commissioned by Infor found only 14 percent of German decision-makers comfortable delegating critical processes entirely to machines, with 43 percent of AI-generated outputs still requiring expert human review.
GOVERNANCE CATCHES UP, SLOWLY
Germany’s regulatory architecture for AI is still being assembled even as the EU AI Act’s obligations phase in. The KI-MIG, or AI Market Surveillance and Innovation Promotion Act, was progressing through parliament in spring 2026 and formally designates the Federal Network Agency, known as BNetzA, as the country’s primary AI market surveillance authority, alongside a new coordination and competence centre. High-risk obligations under the EU AI Act take full effect from August 2026, covering conformity assessments, mandatory risk documentation, and transparency requirements such as labelling AI-generated content and chatbot interactions. For most Mittelstand firms acting as deployers rather than developers of AI systems, the practical burden is manageable, but the definitions matter: a company that substantially modifies a system or markets it under its own name shifts into the far more demanding role of provider.
Public sector use of AI has expanded under Merz’s “Modernisation Agenda,” which targets a 25 percent reduction in bureaucracy costs by 2029 and includes AI-assisted visa reviews and export documentation. The Agentic AI Hub, launched in March 2026 by Wildberger’s ministry, paired seventeen municipalities with ten start-ups across eighteen pilot projects covering tasks such as processing housing benefit applications and naturalisation cases, selected from around 400 start-up and nearly 200 municipal applications. A companion tool, SPARK, uses agentic AI to accelerate planning and approval procedures by checking application files for completeness and flagging issues for human decision-makers, and was released as open source in April 2026. In the judiciary, Baden-Württemberg’s OLGA system assists with case categorisation and Frankfurt’s district court has piloted a tool called Frauke to help draft repetitive judgments, though human oversight remains mandatory throughout.
TALENT, VISAS, AND THE INNOVATION GAP
Germany’s demographic arithmetic remains a genuine constraint on its AI ambitions. The working-age population has shrunk by roughly four million over the past decade, and the country counts more than 1.8 million unfilled vacancies, with acute shortages in engineering and chip manufacturing. The High-Tech Agenda’s talent provisions, unveiled by Merz in November 2025, promise expedited visas, stock-option tax relief, and automatic recognition of foreign degrees for AI, semiconductor, and clean-tech specialists, modelled loosely on France’s Tech Visa. The measures still require enabling legislation, and opposition parties have pushed for safeguards against wage dumping.
On pure innovation metrics, the record is genuinely mixed. Patent output has largely flatlined even as university spin-outs have increased, according to analysis cited by Science|Business. Opposition politicians have criticised the government for allocating just 5 billion euros of a much larger 500 billion euro debt-funded package to research, a fraction that critics argue undersells the rhetoric of an “AI nation.” The coalition agreement between Merz’s Christian Democrats and the Social Democrats invokes that phrase twice within its first hundred lines, and economic policy occupies 80 of its 146 pages, giving AI an unusually prominent place in the governing document even before considering the money behind it.
WHAT “STUCK” ACTUALLY MEANS IN 2026
Unlike France, whose Cour des comptes published a detailed public reckoning of its national AI strategy in November 2025, Germany has not yet produced an equivalent independent audit of its own. What exists instead is a scattered but consistent picture assembled from industry surveys, ministry statements, and comparative EU data: strong and growing public investment, a rapidly improving adoption curve among large firms, a newly credible sovereign AI vendor in the Cohere-Aleph Alpha combination, and a still-real gap between pilot projects and production-grade deployment across the wider economy, especially among smaller firms and public administration.
The word “stuck” is probably too strong for where Germany stands in mid-2026, and too generous for where it stood in mid-2025. Adoption has genuinely accelerated. Compute capacity is genuinely expanding. What has not yet arrived is proof that the scale of public investment, spread across ministries, competence centres, and a six-technology High-Tech Agenda, is translating into the kind of broad-based productivity gain that would let Germany close the market-value gap Wildberger pointed to on stage in Hamburg. That translation, from funded to transformed, is the test the next eighteen months of implementation will need to pass.
Key References and Further Reading
- Artificial Intelligence — BMFTR — Federal Ministry of Research, Technology and Space
- Germany AI Strategy Report — AI Watch — European Commission Joint Research Centre
- Germany AI Strategy — Digital Trade and Data Governance Hub — George Washington University
- Germany Sets Fourfold AI Capacity Target for 2030 — Xinhua
- Digital Minister Wildberger: AI as an Opportunity for Industry — heise online
- Artificial Intelligence 2026 — Germany — Chambers and Partners
- Germany Launches €18 Billion High-Tech Agenda — Science|Business
- Cohere Acquires, Merges With Germany-Based Aleph Alpha — TechCrunch
- Cohere Acquires Aleph Alpha: Deal Born of Sovereignty, Necessity — Futurum Group
- German Mittelstand 2026: AI Boom Meets Strategy Gap — Innobu
- Bitkom AI Study 2026: 41% of Companies Use AI, SMEs Catch Up — MyBusinessFuture
- German Tech Policy Shows New Ambition — Center for European Policy Analysis
- AI Sovereignty for Germany and Europe: Deutsche Telekom Launches AI Factory With NVIDIA — Deutsche Telekom
