AI

Germany's biggest AI missteps: asleep and scattered

5 min read
A half-empty German data center with long rows of server racks, many slots dark and empty, a small German flag on a cabinet, and the Frankfurt skyline at dusk through the windows. Image generated with GPT Image 2
A half-empty German data center with long rows of server racks, many slots dark and empty, a small German flag on a cabinet, and the Frankfurt skyline at dusk through the windows.

TL;DR Too Long; Didn’t read

Germany's AI backlog is not a singular failure but a chain of hesitation and fragmentation: too much hope in individual champions, neglected computing infrastructure, a botched Gigafactory bid, regulation before framework conditions, federal friction, and a left-behind Mittelstand.

Key takeaways

  • Aleph Alpha shows the cluster risk of the champion strategy: much political hope, then pivot, leadership change, and layoffs.
  • In computing infrastructure, a capacity gap of around 50% is looming by 2030 – the joint Gigafactory bid fell apart into three unsuccessful individual applications.
  • The EU AI Act and a national law passed only in 2026 create uncertainty – whether they are the actual brake on innovation remains disputed.
  • While US hyperscalers invest billions, many German providers wait for state guarantees.
  • 43% of medium-sized companies had no concrete AI plans – perhaps the most dangerous gap.

Germany has not failed at a single point in artificial intelligence – it has become entangled over the years in a chain of decisions, the consequences of which are now, in mid-2026, clearly visible. To put it in context: “wrong decision” is always a judgment in hindsight. Much of what is stated here was controversial at the time of the decision and remains so to this day. Therefore, I will try to consider the counter-perspective for each point.

1. Bet everything on a “national champion”

Few stories symbolize German AI policy as much as Aleph Alpha. The start-up, founded in Heidelberg in 2019, was declared by politics and business to be the European answer to OpenAI and raised around half a billion by the end of 2023 – with Schwarz Group, SAP, and Bosch on board, later also Deutsche Bank.

However, a single company cannot replace an entire ecosystem. In September 2024, Aleph Alpha abandoned the construction of its own top language model and became a platform and orchestration company for authorities and corporations. In 2025, founder Jonas Andrulis lost operational leadership, and in early 2026, he left the company entirely; the Schwarz Group became the dominant anchor investor. In January 2026, a wave of layoffs followed, affecting around 50 positions.

Investor Fabian Westerheide pinpointed the design flaw: too much expectation had been projected onto a single company from the beginning – a figurehead to calm the public instead of investing in the breadth of an ecosystem.

Counter-perspective: Aleph Alpha’s pivot to secure, EU-compliant administrative and industrial applications may prove to be economically wiser than a futile arms race against US billion-dollar budgets. So the failure is not necessarily the pivot – but the political narrative attached to it is.

2. Sleeping on infrastructure – and then getting tangled in the application

Computing power is the actual key resource in the age of AI, and here Germany is falling behind. A Deloitte study estimates the impending capacity gap by 2030 at around half of the additional demand; the investment requirement is up to 60 billion euros, while Germany’s market share is already declining in international comparison. High energy prices further exacerbate the problem.

Particularly bitter is the Gigafactory debacle of 2025. In the EU initiative for high-performance AI data centers, SAP, Deutsche Telekom, Ionos, the Schwarz Group, and Siemens originally wanted to join forces. In the end, Germany submitted three separate applications – all of which were unsuccessful. Instead of consolidation: parochial thinking among those who should know better.

This fits with the symbolic withdrawal of Intel from the planned semiconductor plant in Magdeburg – a setback also for chip sovereignty, which is closely linked to the AI location.

Counter-perspective: Germany is not coming away empty-handed. With JAIF in Jülich and HammerHAI in Stuttgart, the country operates two EU-funded AI factories, and for the next Gigafactory round (awarded summer 2026), German consortia are again strongly represented. Some experts consider the “gigantomania” of huge individual data centers to be risky anyway – the Netherlands have consciously moved away from it.

3. Regulation before innovation

Europe aimed to set the global standard with the EU AI Act. The price: from August 2026, obligations for high-risk systems take effect, and uncertainty in the economy is high. Over 40 CEOs of large industrial companies – including Siemens, Airbus, Mercedes-Benz, and ASML – warned in an open letter of competitive disadvantages compared to the USA and China.

Germany has shot itself in the foot: the national implementation law only came in February 2026 – late, with a correspondingly long phase of legal uncertainty for start-ups and SMEs, who did not know which authority would check what and how.

Counter-perspective: There are good arguments that the AI Act is not the actual brake on innovation. Mandatory regulatory sandboxes, clear liability rules, and trust in secure systems can even promote innovation. Critics may confuse cause and effect: the real obstacles – capital, skilled workers, bureaucracy overall – lie deeper.

4. Federal fragmentation instead of a joint effort

Germany indulges in a multitude of its own state strategies alongside the federal strategy, which differ significantly in depth and focus. As early as 2024, experts warned against the “fragmentation in the federal system.” Fragmented responsibilities and lengthy coordination processes delay projects – a structural disadvantage that smaller countries like Luxembourg consistently avoid.

5. Waiting for the state instead of acting entrepreneurially

A cultural pattern runs through it all: while US companies create facts – AWS is investing nearly eight billion euros in a “European Sovereign Cloud” in Potsdam by 2040, Microsoft is building in North Rhine-Westphalia, Google in Hesse – in Germany, the federal government, states, and industry negotiate subsidies, operating-cost guarantees, and state purchase commitments. The state is expected not only to kickstart but also to step in as an anchor customer. Positive exceptions like the Schwarz Group, which took the lead early, or the Telekom AI factory with Nvidia in Munich confirm the rule rather than refute it.

6. Leaving the Mittelstand behind

The perhaps most consequential wrong decision is one of omission: the broad Mittelstand was left to its own devices for too long regarding AI adoption. According to the AI Index Mittelstand (DMB/Salesforce), 43 percent of SMEs had no concrete AI plans, and a supplementary survey found that 68 percent of SMEs had no AI strategy at all – while 91 percent of large companies already consider AI to be business-critical. The gap between pioneers and laggards is growing rapidly, and it is precisely in this gap that Germany’s industrial future is being decided.

Conclusion

The common pattern is not stupidity, but hesitation and fragmentation: too much hope placed on individual lighthouses, too little on an ecosystem; infrastructure addressed too late and disjointedly; rules placed before framework conditions; federal friction instead of consolidation; waiting for state security instead of entrepreneurial courage. The good news: research, talent, and industrial substance are all there. There is still time to change course – but the window is closing.

#AI

← Back to the blog