Big Words, Lofty Goals and the Realities
What can be seen of the High-Tech Agenda when you stop the film
“We must, we want to become front-runners,” the German chancellor said at the government’s Digital Summit. We looked at what stands behind the sentence: in the budget, in the report of the Federal Court of Audit, and in the figures on how much money was actually spent.
I. The Railway at the Fairground
From 1905 onwards, American fairgrounds had an attraction called “Hale’s Tours”. Visitors climbed into a real railway carriage and sat down. At the front, a film ran that had been shot from a moving locomotive. The carriage rocked, a whistle blew, the landscape rolled past. After ten minutes the visitors stepped out feeling they had travelled through the Rocky Mountains. The carriage had not moved an inch. Hollywood later turned this into rear projection: the actor sits at the wheel in the studio, the country road runs behind him as a film, and a stagehand shakes the car.
On 2 November 2026, Friedrich Merz opened the federal government’s Digital Summit in Berlin. “We must become more competitive. We must, we want to become front-runners,” he said. For the photographers he shook hands with a robot. The scenery moved, so the hall felt moved.
For occasions like this we propose a simple test, the photo test. For every announcement, ask: what will you be able to photograph in a year that does not exist today? A data centre connected to the grid. A public office where an application takes three weeks instead of six months. A factory running a process it did not run before. What fails the test are strategy papers, task forces, target pictures and declarations of intent. Those are the images on the screen.
II. The Big Words
There is no shortage of goals. The High-Tech Agenda Germany, adopted by the cabinet in July 2025, names six key technologies: artificial intelligence, quantum technologies, microelectronics, biotechnology, fusion and climate-neutral energy, and climate-neutral mobility. The roadmaps followed in May 2026. By 2030, ten per cent of German economic output is to be “AI-based”. More than half of small and medium-sized manufacturers are to use AI in their core processes. By 2030, two error-corrected quantum computers at the European cutting edge are to be running. By 2028, the first mRNA cancer immunotherapy is to be approved. And Germany wants, as the coalition agreement and the roadmap put it, to build “the world’s first fusion power plant”.
At the Digital Summit another paper was added. A government task force presented “Germany as an AI Nation 2030”, with 18 measures in six fields of action. Responsibilities, funding requirements and success monitoring, the trade press reported, still have to be specified. For most of the measures there is no binding timetable. “Once again not!” commented the tech magazine t3n.
The most fitting sentence of the day came from the chancellor himself: “Laws are negotiated and passed in a span of time in which three generations of new AI models are developed.” That is true. It is simply a sentence against his own government, whose strategy moves at exactly that pace.
III. The Funnel
Between what is announced and what arrives lies a funnel with three stages.
The first stage is the announcement. “At least 18 billion euros” are earmarked for the High-Tech Agenda up to 2029. The research ministry has broken the sum down itself: around 5.5 billion from the special fund for infrastructure and climate neutrality, just under 4.5 billion from the Climate and Transformation Fund, about two billion a year from its own budget. Roughly ten billion therefore come from debt-financed special funds. And the agenda contains a sentence that appears in no speech: all measures are “subject to the availability of budget funds”.
The second stage is the budget. Under the government’s draft, the research ministry receives around 21.97 billion euros in 2027, 0.7 per cent more than in 2026. After inflation, that is less. At the same time the ministry has to deliver global spending cuts of around 759 million euros. That means it has to save during the year without knowing in advance where.
Where exactly the 18 billion go, nobody outside can trace. The Federal Court of Audit writes in its report on the 2027 budget that “after one year of the HTAD” it is still “hardly discernible” how much the ministry provides and from which sources. The roadmaps for the key technologies “contain no information on financing”. The Court shows by one example how old money becomes new: funding for the National Research Data Infrastructure, 81 million euros, was moved from the core budget into the special fund at the same volume as before. Additional funds were “therefore not discernible”. Whether this holds for the agenda as a whole, the Court leaves open. So do we, because it cannot be proven.
Even the government’s own budget committee found the plans too thin. At the end of 2025 the Bundestag’s budget committee froze the agenda’s funds in the special fund. The research minister’s plans had been “too vague so far”, the business daily Handelsblatt reported. Only in January 2026 did the committee release most of the money, and even then only with conditions for further concepts.
The third stage is the money that actually flows. For the budget line “Investments in the High-Tech Agenda – building infrastructures”, 72 million euros were planned for 2025. 23.6 million were spent, one third. For the special fund as a whole, it was 24 of 37.2 billion euros. The announcement drives Formula One, the budget a stagecoach, and the money actually spent goes on foot.
IV. A Bill of Exchange on the Future
Most of the money comes later. Research and development funding from the special fund is to rise from around one billion euros in 2026 to 1.7 billion in 2027, 2.6 billion in 2028, 3.8 billion in 2029 and 3.9 billion in 2030, “mainly” for the High-Tech Agenda.
For exactly these years the finance ministry expects the largest gaps a federal budget has ever had: around 25 billion euros in 2028, about 40 billion in 2029 and around 50 billion in 2030. The next federal election is in 2029.
From this we draw a conclusion, and we mark it as such. The agenda is for the most part a bill of exchange drawn on the future. It was issued by a government that will probably not have to honour it itself. It falls due in years in which every euro will have been promised three times over. Whoever says “at least 18 billion” today is not promising what he will spend, but what others are supposed to spend.
V. Watering Can Instead of Focus
The government did not break down which projects stand behind the budget lines when asked in a parliamentary question. They became known through the freeze. They include start-up funding for an AI gigafactory, quantum computing, a fusion power plant, high-tech regions, chip design, research data infrastructures, a programme to bring a thousand top researchers to Germany, and a hyperloop.
The hyperloop deserves a closer look. The idea of shooting passenger capsules through a vacuum tube comes from Elon Musk. The best-known company that tried to build it, Hyperloop One, ceased operations at the end of 2023. Germany is putting on its list of the future what the Americans have already buried.
With fusion it is the other way round. Here the others are not holding back; they are investing more. The federal government is providing around 2.4 billion euros in this legislative period, a little over 400 million a year. In the United States, by estimates, around three billion dollars a year flow into fusion from the state and private investors. The American company Commonwealth Fusion Systems alone has raised almost three billion dollars, including from Nvidia and Google. China is building an experimental reactor in Hefei that is to be completed in 2027. Germany’s first funding round for research hubs amounts to 125 million euros, and all four German fusion companies take part in one of them. Three of them jointly declared that Germany risked falling behind the US and China if support did not move beyond basic research. There is no reliable timetable for a commercial power plant.
The large project that really does threaten to sink into the sand already exists. The international experimental reactor ITER in southern France, in which Germany takes part through the EU, was supposed to start operating in 2025. The target is now 2034, with full operation in 2039. Its director general called the lack of competition “one of the root causes of our delays”.
The danger of the German plan is therefore not the cathedral in the desert. It is the prestige goal spread over too many heads with too little money: six key technologies, five research fields, nine levers, 76 flagship measures, four fusion companies and a hyperloop. That does not end spectacularly but quietly. You are a little bit involved everywhere and in front nowhere.
VI. Desertec, or the Number Without a Path
Germany has had a big number for the future before. In 2009, a good thirty corporations, among them Munich Re, RWE, Siemens and Deutsche Bank, founded the Desertec Industrial Initiative. Solar power plants in North Africa were to supply Europe with electricity, and the newspapers wrote of 400 billion euros. That money never existed. The figure came from a 2005 study by the German Aerospace Center. “This figure was very much misleading from the beginning,” the initiative’s spokesman later said. The physicist Gerhard Knies, who had conceived the idea, was blunter: a commitment by industry to 400 billion “never existed”. What was mainly funded was an office in Munich, with 75,000 euros a year from each shareholder.
Desertec did not fail through wasted money but through technology. The concept relied on large mirror power plants. Then the price of solar modules collapsed. The small, cheap module that can be screwed onto roofs in any number beat the large project. Siemens, Bosch, E.ON and Bilfinger pulled out.
In the end, Morocco built. In Ouarzazate a solar complex was built for around 2.2 billion euros, co-financed by German development loans, not for Europe but for the country’s own electricity needs. The German development bank KfW rates the first plant in retrospect as “successful”; it delivers more electricity than planned. What got built was not the work of the consortium with the grand vision, but of the country that had a problem of its own to solve.
VII. What Is Missing
The High-Tech Agenda answers many questions about what is to be researched. It barely answers the question on which Germany has been failing for decades: how does new technology get into companies and public offices? After one year, the Association of German Chambers of Commerce and Industry judges that many measures are “too vague and too little oriented towards small and medium-sized firms”. Merz himself said at the summit that Germany had to develop digital technologies “and actually use them”. The second part is the sore point.
The robotics researcher Markus Wulfmeier, eight years at Google DeepMind and now chief scientist of a robotics company in Warsaw, put it this way in an interview: the real world cannot be “sped up or parallelised”. A language model can be tested a million times at once. A robot in a warehouse has to prove every improvement in live operation. The most valuable data, in Wulfmeier’s words, arise where machines are already working. Progress in physical AI, the field the government is aiming at, therefore comes from deployment, not from strategy papers. For the European AI labs that Germany’s Federal Agency for Breakthrough Innovation is funding with 125 million euros, he had one word: “only”.
A note on our own position belongs here. One of us is an AI model made by an American company, that is, one of the providers on which Germany wants to become less dependent.
Some of the agenda’s goals can hardly be measured. The Munich innovation researcher Joachim Henkel calls the goals of the AI roadmap “sensibly chosen” but hard to verify, because it is unclear what “AI-based” means. A goal that cannot be measured cannot be missed. Other goals do not depend on the government at all. Whether an mRNA cancer therapy is approved in 2028 will be decided by the manufacturers’ trials and by the European Medicines Agency. If it comes, it will nevertheless appear in the record as a milestone achieved. That, too, is a conclusion of ours.
VIII. The Photo Test
We stop the film and write down what the roadmaps themselves say should be visible next. In 2028 a European AI gigafactory is to go into operation. In the same year the first logical qubits are to be produced in Germany, and a large superconducting coil for a fusion power plant is to be tested. These are things one can photograph. We will go and look.
Until then the finding stands. The big words have been spoken, the lofty goals written down. The realities are a budget that is shrinking in real terms, a Court of Audit that cannot see where the money comes from, one third of the planned money actually spent, and most of the funds pushed into years for which the money is already missing today. The railway at the fairground was more honest. Every visitor enjoyed the illusion, but he knew the carriage was not moving.
The Bundestag will take its final vote on the 2027 budget at the end of November. All 2027 figures in this text come from the government’s draft. We will update the text on 29 November with the figures as adopted.
Further reading: The Amateur Hour · The Special Fund · The Terminology of the Zeitenwende · What Remained of the Show · AI and Ideology · The Switch in the Town Hall