This year, Germany is spending €1.6 billion on artificial intelligence — more than on almost any other technology. The Federal Ministry of Research, Technology and Space has set aside this sum as part of its AI action plan. Since 2017, the annual budget has increased more than twenty-fold.
According to a survey by the digital association Bitkom, German companies are also keen to invest in AI applications. Around 60% plan to invest at least as much as they did in 2024, while nearly a quarter want to increase their spending on digital tools.
However, digitalisation and sustainability researcher Rainer Rehak from the Weizenbaum Institute warns: “There is currently no business model for AI”.
For big tech companies like Google, Microsoft, or OpenAI, as well as at smaller levels, AI remains an investment game.
‘No return on investment yet’
“It’s all investment money, they’re spending billions and trying to push AI into the market at every turn because they hope that there will be a return on investment at some point,” said Rehak. “But that doesn’t exist at the moment, not anywhere,” warned the scientist.
Only once companies generate profits and financial benefits through AI applications can there be a true “use case,” he added. If this doesn’t happen, “then all these investments will eventually collapse,” predicted the researcher.
In his eyes, companies are currently at a crossroads.
“Market analyses are now asking, is the AI bubble bursting?” Rehak said. More and more experts and analysts fear that investment in AI is progressing too quickly.
Even the CEO of OpenAI, the software company behind ChatGPT, is sceptical about massive spending in the AI sector. According to a report in The Verge magazine, Altman said that the market for artificial intelligence is increasingly becoming a bubble.
So is the AI bubble about to burst?
How much is Germany investing in AI?
In its coalition agreement, the new German government set the goal of “strengthening Germany’s position as a data centre hub and a beacon for Europe”.
“Germany was once a leader in AI research and among the first to publish a national AI strategy,” said Bitkom president Dr. Ralf Wintergerst. “But now we’re lagging behind the US, especially in the field of generative AI.”
Instead of simply compiling and analysing data, generative artificial intelligence can create new content. But this requires resources.
Requests put to ChatGPT and other large-scale computing tasks are implemented in high-performance data centres. But Germany does not have enough of them, according to a study by the eco Internet Industry Association.
How much computing power does Germany have available?
According to the eco study, Germany’s computing capacity could be increased to up to 3.7 gigawatts by 2030, a 50% increase within five years.
However, demand from industry is set to be three to five times higher. Up to 12 gigawatts of power could be needed, equivalent to the output of at least ten nuclear reactors.
The US already has 20 times Germany’s current computing capacity, with projections showing even faster growth for the US by 2030.
The consulting firm Deloitte has warned that at the current rate of expansion, there will be a capacity gap of around 50% by 2030, meaning “massive additional investments” will be needed to meet demand.
So should Germany pour even more money into AI, and just keep waiting for profits?
‘Understanding AI as a networked ecosystem’
“Germany and Europe are paying today for the IT gigantism of the 1990s and 2000s,” said Professor Oliver Thomas of Osnabrück University, founder of the consulting firm Strategion GmbH. His own companies are already betting on AI.
While providers of large AI models still struggle to monetise them, some businesses can already create value from AI.
Even so, Europe won’t reach the scale of US tech giants like Microsoft and Amazon, Thomas argued, as these firms saw the potential of AI two decades ago and invested early.
He has therefore adapted his own company’s approach.
“We have to be faster,” said Thomas. “So research and development, implementation and commercialisation take place in parallel”, instead of sequentially.
He continued: “We need to assess as early as possible what the leap into reality will look like. In a figurative sense, AI now has to go where it belongs, namely where our appreciation takes place in Germany: among SMEs and hidden champions.”
The so-called hidden champions are relatively unknown larger companies, some of which are market leaders in their sector.
Thomas is also appealing to politicians. According to him, the German government’s AI action plan is fundamentally correct, but “researchers, universities and institutions must take more responsibility for ensuring that knowledge transfer works”.
Profits can only be realised, he argued, when research — the previous focus of federal AI funding — “is no longer just theoretical, but is directly put into practice”.
Technical expertise made in Germany?
“AI was in a deep sleep,” said Thomas on the situation in Germany.
It was only through applications such as ChatGPT that the general public became aware of the benefits of artificial intelligence. Interest in it then increased, both privately and commercially.
Germany must now be prepared to make up for lost potential, he said.
“Think of computer technology, the first AI processes, the MP3 standard, virtual reality, augmented reality. These are all topics that we have actually always researched extensively, but which have been commercialised abroad. That’s why many AI models don’t come from Germany.”
Thomas’ solution is to focus on applying AI from abroad, rather than focusing too heavily on ‘made in Germany’ technology. At the same time, he warns that digital sovereignty must not be lost, and it is the responsibility of politicians “to set up programmes to counteract this, such as the action plan”.
AI colleagues and virtual assistants in the workplace
Looking ahead, Thomas sees new types of digital companies emerging. He predicts that even small teams could generate millions in revenue with the help of AI colleagues and permanent AI assistants.
An AI-managed archive, for example, can retrieve the right document in a matter of seconds. Digital assistants pre-sort emails, organise workflows, and document the necessary steps, for example in fields like tax consultancy or auditing.
AI workers could become so integrated into the workforce that they would show up as part of the official staff structure.
