What Every European Must Know

Ten statistics that will decide Europe’s future

Every year drowns us in an ocean of headlines. Most of them will not matter in ten years, but a few of them will.

For Europe, a small set of statistics influence what our continent will look like in 2050: how wealthy we are, who lives here, how much we can care for one another, and what kind of politics we get.

Over the past year I kept coming back to these numbers. Each felt like the piece of a puzzle, the ragged outline of one fitting into the uneven gaps of another, and together they began to form a picture. This post is my attempt to assemble those pieces: Ten areas and ten sets of statistics that every European must know.

1. Growth

Since 2008, EU GDP has stagnated, totalling ca. $20tn in 2025, while China and the US have grown their GDP more steadily over time.

The IMF expects the Euro area to grow at around 1% per year in 2025–26, the US at roughly 2%, China at just above 4%, with global growth a little above 3%. In other words, unless something changes, Europe will keep shrinking as a slice of the global pie.

2. AI

US GDP growth in H1 2025 was driven almost entirely by investment in AI infrastructure, that is: data centres, as Harvard economist Jason Furman pointed out:

Without this heavy investment, US GDP would have been close to stagnation, much nearer to the Europe’s performance, which is roughly half as strong and not yet underpinned by comparable AI infrastructure spending.

The point, however, is that investment in AI infrastructure is not just a statistical trick to boost an otherwise struggling US economy. It is a high-conviction investment in future productivity. As AI diffuses through the economy, it is expected to lift future economic growth. To grow an advanced economy today is increasingly to win the AI transformation. Economists at the University of Pennsylvania, for example, project AI-driven GDP gains of around 1.5% by 2035 and about 3% by 2055.

3. Chips

Value creation from AI is significantly driven by AI chips, which could capture up to 50% of total value created by the AI technology stack over time (McKinsey).

Market share for logic chip production, i.e. the chips that power modern compute, is squarely concentrated in Asia and the US. The few notable exceptions are small number of European companies, represented in the “Other” category in the graph below (Our World in Data).

While Europe plays only a minor role in total AI chip value creation, it does have a few important players. As the sole manufacturer of complex EUV lithography systems the Dutch company ASML has significant market power and is, as of December 2025, the only remaining European company in the list of top 25 global companies by market capitalisation. The German companies Zeiss, which produces highly complex EUV optical lenses, and Trumpf, the producer of high-power laser systems, also deserve a mention.

If Europe wants real sovereignty in AI, we will have to move beyond being a niche supplier of tools and components and invest at scale in chip design and manufacturing. Otherwise we will remain structurally dependent on foreign fabs for the hardware that runs our future economy.

4. Models

The US and China make up well over 100 large scale AI models (models whose training compute exceeded 10^33 FLOPs) and are on a steep forward trajectory. The UK, Germany, France, Switzerland, Ireland, Belgium and Finland combined are getting to ca. 88 large scale AI-systems with a weaker trajectory (Epoch AI & Our World in Data).

Today, most AI usage flows through non-European models. The bar chart below shows total AI token usage over time. The bulk of usage continues to go to closed models from companies like Anthropic, OpenAI and Google, without a single strong European contender. Open source models are gaining share, but this segment is currently led by Chinese models such as DeepSeek. European AI models sit inside the “rest of world” category, with Mistral as the only notable player (OpenRouter).

If Europe wants to play a leading role in the world’s intelligence transformation, we need to build and back our own winning foundation models rather than relying almost entirely on US and Chinese systems.

5. Investment

Training AI foundation models requires immense amounts of capital. Investments that boost innovation have grown less strongly in Europe than in the United States since the early 2000s. After the financial crisis in 2008, investment in business Research and Development between the EU and the United States began to diverge significantly (OECD).

Similarly, the US plays an extremely dominant role in VC funding of AI startups: in the first half of 2025, the US accounted for 97% of global venture capital AI deal value (the total $ invested in AI companies) and 62% of AI deal volume (the number of AI VC deals), EMEA accounted for 23% of volume but just 2% of deal value (EY).

Nonetheless, a number of European startups have succeeded in raising the hundreds of millions in funding required to build a large company in a very competitive market.

However, of 147 unicorns that were founded in Europe between 2008 and 2021, 40 moved their headquarters abroad, the bulk of them to the United States (European Commission).

If Europe wants world-leading AI companies to be built here and stay here, we need a concentrated effort to match global capital at billion-Euro-scale and to tackle our structural frictions, above all the severe market fragmentation that still pushes many of our best firms abroad. Initiatives like EU-INC are a first step in the right direction, but they need bold implementation rather than timid compromise. We should focus on fewer things, but pursue those much more aggressively.

6. Adoption

When it comes to AI adoption, European countries like Germany and the Netherlands, are very much punching above their weight. AI inference provider OpenRouter analysed over 100 trillion tokens of AI usage. The result: Germany accounts for 7.5 % of LLM tokens processed via the platform, while generating only around 4.2 % of global GDP.

Crudely summarised, Germany’s AI footprint is therefore almost twice as large as its economic footprint. The US is responsible for a much higher share of global LLM tokens (around 47 %), but relative to its share of world GDP its AI usage intensity is roughly similar to Germany’s. This pattern is echoed in data published by OpenAI, which cites Germany as its third-largest global market by paid subscribers.

AI already significantly boosts individual worker productivity, with strong increases in AI usage reported across various work functions, from marketing, to software development or research. In a global study among 2,430 researchers, ExplanAItions found that 84% of researchers are using AI tools for any aspect of their work, up from 57% a year earlier.

Similarly, a study of 153 engineers working at AI foundation model company Anthropic self-reported using Claude in 60% of their work and achieving a 50% productivity boost, a 2-3x increase from this time last year (Anthropic).

Similar to the previous digital divide, an AI divide is increasingly decoupling companies that adopt from those that do not. AI adoption boost revenues and cuts costs, enabling companies to reinvest profits into even stronger future growth, creating a growth flywheel that non-adopters struggle to compete with (BCG).

In many cases, cost cutting is driven at least partly by headcount reductions. While it is still too early for truly dependable studies, consultancies like (McKinsey) and leading economists estimate that up to 30 % of US jobs could be automated by AI by 2030, echoing the warnings of many AI leaders such as Dario Amodei. We do not yet know precisely how many people will be affected by AI automation, or how quickly, but we do know that those who neither build AI systems nor adopt them well will almost certainly be left behind.

If we double down on our early lead in AI adoption, we can turn productivity gains into broadly shared prosperity, reinvesting them in skills, retraining and new forms of work. If done well, automation becomes a driver of higher living standards and renewed social mobility rather than a source of insecurity.

7. Electricity

Any developed nation wanting to grow its economy must step up to lead in AI, and AI inference in data centres requires significant electricity resources.

Electricity prices in Europe are twice as expensive as in the US or China. Experts estimate that 30–50 percent of a data centre’s operating costs are electricity, so these costs matter enormously.

Electricity generation also differs by country, with the US and China leading Europe by 63% and 274% higher electricity generation. Most notably, both countries, and most impressively China, are showing a stronger growth rate in expanding electricity generation capacity, while European electricity demand and energy supply have remained stagnant since 2007. China is taking the lead largely due to heavy investment in renewable energy.

In 2024, US data centres used around 183 terawatt-hours (TWh) of electricity, compared with about 70 TWh in Europe. In line with global trends, European data centre demand is projected to rise to roughly 115 TWh by 2030 (IEA).

Across Europe, data centres are already hitting hard constraints: power systems struggle to absorb large, concentrated new loads due to limited reliable generation, strained urban grids and years-long wait times for grid connections, equipment, land and skilled labour. As a result, many projects face delays of two to ten years and mounting political pressure to ensure that data centre growth does not drive up electricity prices for households and industry.

Without very large new infrastructure investment, this demand surge will put energy systems under extreme pressure, which is why some tech CEOs now float the idea of putting data centres in space to tap constant, unobstructed solar power. If Europe wants to meet future AI-driven electricity demand, it will have to invest at scale in energy infrastructure and deregulation or see growth strangled by severe bottlenecks.

8. Migration

Next to capital and deregulation, Europe’s innovation strength is first and foremost decided by its people: who is born, who leaves, and who arrives. The below data focuses on Germany, though countries like the UK are seeing similar trends.

On the whole, Germany still has an overall positive net migration balance. In 2024, Germany gained around 430,000 people in total – but this due to immigration by non-German citizens (Statistisches Bundesamt). By contrast, among German citizens, more people are leaving than returning to the country.

This mobility is not evenly distributed. Highly educated people are more likely to move for work or opportunity - around 76% of Germans leaving have a university degree (BIB).

According to Henley & Partners, in 2025, for the first time since they started collecting data on the matter, more millionaires are leaving Germany than arriving; they estimate roughly 400 people and about €2 billion of capital leaving. It is a small number in absolute terms and a crude proxy for opportunity loss, but nonetheless an important signal.

When skilled people leave and do not return, the country loses not just tax revenue, but a significant part of its future potential: initiative, company formation and leadership. We are seeing the early signs of Germany turning into a brain drain, rather than a brain gain country, as we have prided ourselves in being for so many decades.

Germany’s future depends both on persuading skilled citizens to stay and making Europe one of the most attractive places in the world for talent to move to and build a life. Many of the doctors, nurses, engineers, founders and researchers who keep our society running were not born here, and in the years ahead our prosperity will hinge even more on their decision to come and stay.

9. Birth Rates

Fertility trends add another puzzle piece. Birth rates across Western countries have been declining for decades. Germany’s total fertility rate of ca. 1,35 children per woman ranks among the lower half of birth rates in EU countries and compares to a total fertility rate of 1,6 children for women in the US, as well as 1,0 children for women in China. This contrasts to fertility rates around 3-5 for women in the Global South.

Given the continued population growth in the Global South, with Nigeria and Pakistan each projected to have roughly the same population as the EU by 2050, migration pressures on Europe’s borders are likely to persist.

Women who move to Germany from countries with higher fertility rates have, on average, more children (2,06) than women without a migration background (1,26).

(summarised birth rate of German (light blue line) and foreign women (dark blue line), 1991-2024)

As a result, ca. 43% of children under five in Germany now have a migration background, compared with roughly 30% of the overall population, and that share will rise (Statistisches Bundesamt).

The demographic composition of European societies is changing and will continue to change; the question is not whether this happens, but how well we manage it. If we approach migration with clear rules, serious investment in integration and a confident and expansive sense of who we are, demographic change can become a source of dynamism rather than division.

10. Politics

Demography and migration do not happen in a political vacuum. Across Europe, far-right parties have been gaining ground for years. In the three largest EU countries, they are now polling in first place. Economic stagnation, a sense of cultural loss and national security fears all feed into their narrative.

The world’s most powerful country is fanning these flames. In its recently published National Security Strategy, the US administration wrote that “the growing influence of patriotic European parties indeed gives cause for great optimism” because of the “real and [.] stark prospect of civilizational erasure.”

As a German with a migration background, I reject both the cruelty and hatred of much far-right migration politics and the comforting story that concerns around migration are illegitimate.

Welcoming people of different nationalities is not civilisational erasure. It is difficult, but achievable, societal change. The statistics show that this change is coming, whether right-wing voters like it or not. To get back to cohesive ground, we need to update society’s view of what it means to belong to a country, shifting from a zero-sum view to a positive-sum one in which different cultural strands make society richer rather than threatening it.

Where do we go from here?

Taken together, What Every European Must Know is uncomfortable reading. The data underlines that Europe is losing economic weight, constrained on energy, behind on chips and models, facing demographic headwinds and tearing at the seams politically. On average, the story is not good, and several of these trends negatively reinforce one another in ways that make things worse.

Let’s focus on the silver linings.

1. Europe is not a statistic

While we need to take the statistics seriously, we are not a statistic. Summary statistics only allow limited conclusions about underlying individual data points. From “Swedes are taller on average than Italians” we cannot conclude that any particular Swede will be taller than any particular Italian. Some Italians will indeed be taller than some Swedes, despite the difference in averages.

When pointing out that Europe is struggling in various critical dimensions, we cannot automatically deduce that therefore, it is impossible possible to achieve greatness in Europe. That is why we see great companies emerge even from the most unlikely of places. Greatness always has a way of defying the odds.

Much of the Euro-bashing happening on social media makes exactly this mistake of deduction: “European startups receive less funding than startups based in the Bay Area, hence it is impossible to build a company from Europe and a relocation to the US is the only viable option.” Wrong. In fact, there are plenty of successful AI companies that show that you can build thriving businesses from Europe. There are just many more in the US and we need to change that.

2. A growth mindset for Europe

Pessimistic Euro-bashing is what a fixed mindset looks like in practice. A fixed mindset treats abilities and circumstances as largely innate and unchangeable; a growth mindset assumes they can be developed through effort and persistence.

Those who look at Europe’s summary statistics and conclude that the continent is utterly lost, that our economy is doomed and that our culture will get erased, are operating from a fixed mindset. A growth-mindset is to see the same numbers, accept that the circumstances are difficult, and still believe that greatness is achievable, then act accordingly. The more people adopt a growth mindset vis-à-vis Europe, the more today’s uncomfortable statistics become starting points to act on rather than final verdicts to accept.

3. Europe is not beyond fixing

As the saying goes: a small change in slope can make up for a lot of y-intercept. Even small shifts in direction now will compound into very different conditions a few decades from today. Europe is not unfixable in ten years. And so in the causal chain of our current demise also lies a solution:

While growth is no guarantee of social cohesion, its absence is a recipe for social disaster. In an economy that falters, or worse, decouples from the two economies we have long stood beside in the global marketplace, Europeans who feel they are falling behind eventually turn economic strain into cultural and political strain. If we want any chance of improving social cohesion and Europe’s political climate, we have to fix the economy. That means stepping up in AI and entering a global competition that is getting fiercer by the hour. Our early lead in AI adoption is a rare and important signal that we can lead if we are determined to do so.

In the end, the statistics that scare us today are the sum of individual choices made yesterday. The statistics our children inherit tomorrow will be shaped by what we choose to do now.