The Built-in Customer

The debate about innovation finance almost always asks the wrong question. Who pays? How much? In what form — debt, equity, state budget? The decisive question is a different one: Who buys? Europe's failure at innovation is not a financing problem. It is a customer problem.

I. The Number That Explains Everything

In 2024, China overtook the United States in research and development spending for the first time in history. China spent $785.9 billion, the United States $781.8 billion. The headline wrote itself: the West is losing the innovation race. The conclusions drawn from this focused almost exclusively on money — on subsidy rates, tax incentives, public investment programs.

But the number itself is the least of the problem. More relevant is what lies behind it. China is growing its R&D at nine percent annually; Germany at 0.8 percent. Israel spends 6.3 percent of its GDP on research, South Korea 5.0 percent, the EU 2.1 percent on average. These are not coincidences. They are structures. And behind each structure lies a different answer to the question: who takes what is produced?

II. The American Model: The State as Soldier and Risk-Taker

DARPA — the Defense Advanced Research Projects Agency — was founded in 1958, in the shock of Sputnik's launch. The Soviet Union had overtaken the United States in space. The military needed a response, and needed it fast. What emerged was not a funding program. It was a new organizational form: around a hundred program managers with broad autonomy, a budget of roughly three billion dollars, and an explicit license to fail. The internet, GPS, stealth technology — all have their roots here.

The decisive point is not the budget, and not the autonomy of the program managers. It is what stands behind them: the US Department of Defense with a total budget of over $190 billion annually for research, development, and procurement. DARPA develops. The military buys. The customer is built in. It is guaranteed. It carries the market risk that deters every private investor in early technology phases.

This mechanism is the true core of the American innovation system — not the venture capital market of Silicon Valley, not the universities, not the tax code. The entire civilian innovation economy of the United States sits on a militarily financed foundation. The internet was originally a military communications network. GPS was designed to guide missiles. The first generation of high-performance computers was built on commission from the nuclear agencies. The built-in large customer was the Pentagon — and it still is.

III. The Chinese Model: The State as Planner and Buyer

China has developed a different version of the same principle. Those who wish to understand why "Made in China 2025" has worked — why China today dominates global markets in renewable energy, electric vehicles, and is pushing hard in semiconductors — must think not primarily about state subsidies. They must think about the Chinese state as buyer.

China buys its own electric cars. China buys its own solar panels. State banks finance scaling, state procurement agencies secure the first market, state standards exclude competition until the domestic industry is competitive. This is not a free market. It is also not a pure planned economy. It is state capitalism with a built-in customer — and that mechanism is faster and more brutally effective than any subsidy program.

Financing flows through state banks, through the state budget, and — what is rarely stated openly — also through the People's Bank of China, which directs capital into strategic sectors through targeted credit instruments. Money creation as industrial policy is not a Western concept. In China it is normal practice. The inflationary consequences are absorbed through export surpluses and currency management. The model has costs. But it also has results — and the results are visible.

IV. The Korean and Israeli Models: Two Variants Without the State as Primary Customer

South Korea leads the world in research intensity — five percent of GDP, financed overwhelmingly by corporations. Samsung, Hyundai, LG. These conglomerates are both developers and first customers of their own innovations. They build the machine, buy the machine, export the result. The built-in large customer is not a state but a vertically integrated corporate structure that holds innovation from basic research through mass production in a single hand. State tax reforms have promoted this model — but did not invent it.

Israel has taken a different path. The military — specifically units like Unit 8200 — functions as a state-funded training apparatus and first customer for security technology. Those who develop and deploy systems there leave with experience, networks, and a technology that has already proven itself in an extreme environment. International venture capital comes afterward — and it comes because the product is already proven. Israel does not print money for innovation. It produces proof that something works — with state money disguised as a military budget.

V. The European Model: The Gap No Program Fills

The European Union spends 2.1 percent of its GDP on research — below the OECD average of 2.7 percent, far below the US, Korea, or Israel. More important than this number is the structure behind it: roughly half of private R&D spending in the EU comes from mature industries — pharmaceuticals and automotive. The EU is trapped in the middle technology trap: it is good at what it has always done, and has no structural mechanism that forces it out of this position.

Germany established SPRIN-D in 2019 as a kind of civilian DARPA. The comparison is flattering and false. DARPA has a customer with a $190 billion annual budget. SPRIN-D has an evaluation committee. The difference is not one of degree. It is categorical. What makes DARPA powerful is not its budget of three billion dollars — that is comparatively modest. It is the certainty that what is developed will also be bought. That certainty does not exist at SPRIN-D.

There is an additional structural flaw that is rarely named so clearly: the cost of failure in Germany is prohibitively high. A study published by the Ifo Institute in January 2026 systematically surveyed restructuring costs in various countries for the first time. In Germany, the average restructuring cost is 31 gross monthly salaries per employee made redundant. In the United States, it is seven. Innovation requires taking risks — and risks include failure. A system that makes failure this expensive penalizes innovation structurally. It is no coincidence that Germany has not produced a Google, an Amazon, or an OpenAI. It is systemic logic.

VI. The Special Fund as Symptom

Against this backdrop, the German Sondervermögen — the so-called "special fund" — is particularly revealing, not as an isolated case but as a symptom. Of the 24 billion euros in new debt taken on in 2025 and labeled as investment capital, approximately 95 percent ended up in consumptive expenditures, according to Ifo Institute analysis: energy costs for existing companies, social transfers, ongoing state expenses. For the "HighTech Agenda Deutschland," 4.5 billion euros were earmarked; for energy costs of existing firms, around 30 billion.

This is not misallocation. It is the natural consequence of a system without a built-in customer for innovation. Those who give money without knowing who will buy the result end up buying security — the security of what already exists, of existing jobs, of existing industries. That is rational. It is also the precise opposite of innovation finance.

The decisive difference between the American and the European model is not the scale of spending. It is the question of whether someone is waiting at the end of the process who needs and buys the result. Without that someone, every innovation program is the subsidization of hope.

VII. What This Means

Three financing models have proved structurally successful: the military procurement model of the United States, the state planning model of China, and the corporate verticalization model of South Korea. All three share a common feature: they guarantee the innovator a buyer before the innovation is finished. The risk is not eliminated — it is shifted. From the developer to whoever has an interest in the result and is large enough to carry that risk.

Europe has none of these models. It has programs, subsidy rates, evaluation committees, and reporting requirements. It has Horizon Europe with an annual budget roughly equivalent to what the US Department of Defense spends on research and development in two weeks. It has declarations of intent — from Macron's DARPA speech at the Sorbonne in 2017 to the summit conclusions of the following decade.

What it does not have is a mechanism that connects innovation to guaranteed demand. Without that mechanism, innovation in Europe remains what it structurally is: a venture that the public co-finances without buying the results, and that the market cannot fully finance because it cannot wait for the product to mature.

The debate about debt versus equity versus money creation is therefore secondary. It is the debate about the taps while the barrel is missing. As long as Europe has no built-in large customer — no mechanism that guarantees bridging the gap between development and first market — it will remain in this structural trap. Not for lack of intelligence. Not for lack of money. For lack of someone who buys.