By Nicole Junkermann, founder of NJF Holdings
Earlier this year, Europe was given a useful glimpse of what dependence on artificial intelligence might actually mean.
When access to Anthropic’s latest AI model was briefly restricted outside the United States, the debate around European AI sovereignty suddenly became much easier to understand. What happens if a European company, hospital or government department builds important systems around an AI model and access to it changes? What happens if the decision isn’t theirs to make?
The instinctive answer is that Europe needs to build its own version of OpenAI or Anthropic.
I’m not convinced that’s the right lesson.
As an investor in technology and artificial intelligence, I believe Europe should certainly invest more in AI. It needs compute, energy, research, infrastructure and European AI companies capable of competing globally. There are strong economic reasons for doing all of those things, quite apart from the question of technological sovereignty.
But AI sovereignty and AI self-sufficiency aren’t the same thing.
Modern technology is built across borders. AI chips can be designed in one country, manufactured in another, installed in data centres elsewhere and used to train models developed by researchers recruited from around the world. Investment capital crosses those borders too. Trying to make every part of the artificial intelligence supply chain European would be enormously expensive and, in some cases, probably impossible.
Nor should that necessarily be the ambition.
The more useful question is what Europe needs to control in order to retain the freedom to choose.
Europe already depends on other countries for an enormous number of goods, technologies and services. That’s part of how a global economy works. Dependence becomes dangerous when there’s no credible alternative.
Artificial intelligence makes that distinction particularly important because AI models are becoming infrastructure surprisingly quickly. Businesses are building them into products and workflows. Scientists are using AI for research. Governments are exploring its use in public services. AI is beginning to sit underneath things we rely on without necessarily being visible to the people using them.
Once that happens, changing provider becomes harder.
If a business has spent years building processes, data and products around a particular AI model, losing access isn’t the same as changing an office software subscription. The technology becomes part of how the organisation itself operates.
Europe should therefore worry about concentration in the global AI market. But it should be precise about the problem it is trying to solve.
The goal of European AI sovereignty shouldn’t be to remove every foreign dependency. It should be to make sure no single dependency becomes a point of failure.
That means preserving choice.
There is another part of the AI sovereignty discussion that receives less attention.
When we talk about Europe’s AI capabilities, we tend to talk about physical and technological assets: AI chips, data centres, frontier models, energy and capital. All matter. But technological sovereignty also depends on people.
A country can own enormous amounts of computing infrastructure and still be dependent if it lacks the people who understand how to build with it, improve it and question it.
This is central to the Human Code, the framework I developed at NJF Holdings to think about the relationship between technological progress and human capability.
For me, the important question around artificial intelligence has never simply been what machines become capable of doing. It’s what happens to human capability as those machines become more powerful.
AI should make people more capable, rather than gradually removing their ability to act without it. The same principle applies to businesses, institutions and countries.
Europe therefore needs AI researchers who can work at the frontier. But it also needs engineers who can turn models into useful products, founders willing to build European AI companies, experienced technology investors prepared to finance their growth and customers willing to buy from young businesses.
It needs AI expertise inside government as well. States can’t make informed decisions about systems they don’t sufficiently understand.
Human capital is part of AI sovereignty. It may prove to be its most durable form.
There is a tendency in technology to assume that the biggest platform captures all the value.
Sometimes it does. But not always.
Europe may or may not produce the world’s most powerful general-purpose AI model. European artificial intelligence companies should certainly compete for that position. But Europe’s economic future doesn’t depend on winning that single contest.
There are many layers between a frontier AI model and the point where artificial intelligence actually changes an industry.
Europe has deep capabilities in advanced manufacturing, pharmaceuticals, healthcare, engineering, science, financial services and defence technology. These industries contain decades of expertise, specialist data and institutional knowledge that can’t simply be recreated by training a larger language model.
As a technology investor, I think this is where the European AI opportunity becomes particularly interesting.
The next generation of important AI companies won’t all be frontier model companies. Many will combine powerful artificial intelligence with deep knowledge of a particular industry or problem. They’ll have to work in complex physical environments, understand regulation, earn trust and demonstrate that their technology solves something valuable.
Europe has plenty to build on here.
European investment capital should follow those genuine advantages rather than assume that success requires recreating every American technology company.
Thinking about AI sovereignty as resilience rather than isolation also changes the investment priorities.
Europe needs greater AI computing capacity and reliable energy infrastructure to support it. It needs to retain world-class artificial intelligence researchers and attract more of them. Europe should support open AI models and make it easier for companies to build across different technologies rather than becoming locked into a single provider.
It also needs investment capital willing to stay with European technology companies as they grow.
Through NJF Holdings and our technology investment activities, I’ve seen how Europe has become much better at creating ambitious technology businesses. But it is still too common for successful European companies to look elsewhere when they require very large amounts of growth capital.
If AI sovereignty matters, ownership and financing deserve a place in the discussion alongside research and invention.
None of this requires Europe to shut itself off from American AI technology. That would be a strange response to a concern about choice.
OpenAI, Anthropic, Google and the companies that follow them should remain part of Europe’s AI economy. European businesses should use the best artificial intelligence available to them.
They just shouldn’t have only one door through which they can walk.
We’re still early enough in the development of artificial intelligence that many of these dependencies haven’t hardened.
That won’t remain true forever.
Decisions being made now about European AI infrastructure, investment, education and procurement will help determine which technologies European businesses and institutions depend on a decade from now. Some of those decisions will be difficult and expensive to reverse.
This is why the debate about European AI sovereignty matters, even if its definition needs to become broader.
The goal shouldn’t be a European flag on every AI chip, cloud platform or frontier model. Nor should Europe measure its success by whether it has built an exact equivalent of whatever is leading in Silicon Valley.
A stronger test is whether European businesses and governments can choose between AI technologies and change when circumstances change. Whether they have the expertise to understand the systems they rely on. Whether European founders can build important AI companies here and find the investment capital to keep growing them. And whether the loss of access to one technology creates disruption rather than paralysis.
That is a form of AI sovereignty worth investing in.
Because in the age of artificial intelligence, independence won’t mean building everything ourselves. It will mean retaining the knowledge, infrastructure and alternatives that allow us to choose.
Nicole Junkermann is an international entrepreneur and investor and the founder of NJF Holdings. Through NJF Capital and the wider NJF Holdings investment platform, she has invested in technology businesses across artificial intelligence, cybersecurity, healthtech and other areas of emerging technology. Her Human Code framework explores how technological progress can strengthen human capability, resilience and agency.