Europe is at a crossroads. The advent of AI is the most profound technological leap since the harnessing of electricity, promising to rewire our economies, redefine our defense capabilities, and reshape the fabric of society. Yet, in the face of this epochal shift, a dangerous narrative is taking root: that Europe should simply accept a secondary role, renting its cognitive infrastructure from American or Asian hyperscalers because it is "easier", "cheaper" or “because we cannot match their capabilities”. All wrong assumptions.
This is not a strategy; it is a capitulation. This is a long game, and leading on AI requires the hard, unglamorous work of developing frontier models, because true understanding comes only from doing and pushing our limits. Progress in this field is strictly sequential, so you cannot abandon frontier research now and expect to miraculously possess the capability to build it later. Building frontier models gives us the expertise to understand their limits, challenge their design and decide how they should evolve. Without that expertise, even our choices are shaped by someone else. If we outsource the engine of our future, we outsource our destiny. We can either own our future, or we will be owned by it.
Europe owes a profound duty to its citizens and future generations. We must step up, unite across the EU27, and build the sovereign infrastructure our businesses and economy demand. Sovereignty does not require isolation. It requires the freedom to choose our partners, challenge their terms and build alternatives when our interests diverge. If we fail, we risk descending into a new digital feudalism, where the rules of our society are dictated by AI models trained in Silicon Valley or Shenzhen, reflecting values that are not our own. We cannot choose the convenient path when our fundamental autonomy is at stake.
The paradox is that we already possess the resources to succeed. Europe’s linguistic diversity and deep industrial expertise give us valuable foundations for training models. Critics often claim that European compute infrastructure is inadequate, but we have seen firsthand what is possible. At Domyn, we have proven that Europe can compete at the highest levels. European supercomputing infrastructure already provides the capacity to train frontier AI models. We are drawing on that infrastructure, including JUPITER, to develop Europa, a fully open European frontier model. Domyn Edge, trained from scratch on Italy’s Leonardo supercomputer, is the best-performing fully open model of its size. By sharing its weights, training data, checkpoints and code through a staged release, we enable others to build on that progress. These achievements show why European AI is worth investing in, and why we must make full use of the infrastructure and expertise already available.
Yes, we need more compute, and that capacity is rapidly arriving with the rollout of European AI factories and gigafactories. We must draw on the infrastructure distributed across Europe to train models, test new approaches and advance our capabilities. Imagine what we could achieve with purpose-built, optimized infrastructure. This is precisely why we are aggressively expanding, investing heavily to build the most powerful AI supercomputer in Europe. Specialized, domain-specific models meet immediate enterprise needs, while frontier research expands what those models will be capable of tomorrow. Europe needs both. Building applications while leaving the underlying intelligence in other hands leaves our economy dependent on technology we do not control. That progress must translate into better software: fixing the unreliable systems businesses depend on and accelerating discovery across industries.
Building the technology is only part of the challenge. European businesses must also be willing to buy it. Too often, choosing an established Silicon Valley supplier is treated as the safer decision. Giving European AI companies a route to scale requires customers, not just public funding.
This commitment is doubly critical now, as the current paradigm of digital silicon-based computing approaches its physical limits. The next step change in AI capabilities may come from more than incrementally better GPUs, including a shift toward analog computing approaches such as photonics, which uses light to process data. This transition promises significant gains in speed and energy efficiency, and crucially, it’ is an area where Europe holds a decisive competitive advantage. To shape that future, we must build the expertise today, through frontier research and investment in the technology available now. To abandon frontier research today is to forfeit our leadership in the AI of tomorrow.
Furthermore, our vision extends beyond raw compute. By building decentralized AI infrastructure, we democratize sovereignty, ensuring that the burdens and the massive benefits of AI are shared equitably across the continent, rather than concentrated in the hands of a few monopolistic entities. That means giving more businesses, universities and public institutions the capacity to build and govern their own AI. Sovereignty must put control in the hands of the institutions that depend on it.
Finally, we must recognize that Europe’s unique regulatory environment is not a hindrance, but our strongest asset. Our commitment to democracy, the rule of law, and human rights provides the bedrock for trustworthy technology. The AI Act is frequently mischaracterized as a burden; in reality, it is a formidable competitive advantage. It mandates the openness, safety, transparency, and protections that enterprises and citizens increasingly demand. But rules alone will not secure Europe’s position. They must be clear, workable and matched by investment in the capability to build.
Europe must vigorously defend the principles that make it unique. Our regulatory framework and our emerging sovereign infrastructure provide a powerful third-mover advantage. We should not abandon those principles to satisfy competitors who benefit from our dependence. We should back them with action: expand access to compute, sustain investment in frontier research and give European AI companies a route to market through public procurement and private sector adoption. This must be our guiding light, the foundation upon which we build a more just, prosperous, and fair digital world, not only for Europe, but as a beacon for the rest of the globe. The hard work has already begun. Now, we must see it through.
Frequently Asked Questions
