The Network Challenges Europe Could Face in Its Sovereign AI Push
Artificial intelligence is a two-horse race between the United States and China, and not one that Europe is particularly comfortable with. Amid rising geopolitical tensions, concern surrounds the weaponization of technological dependence through trade, while data protection – an area the European Union leads – remains a worry. With ministers
describing
the need to break this reliance on foreign-controlled systems as a “matter of national survival”, there’s a growing push to achieve AI sovereignty.
Aiming to develop, deploy, and support AI capabilities closer to home, the EU recently announced its EURO-3C project. This will see €75 million invested to create a federated network of nodes that will operate across borders, bringing cloud services closer to European end users and reducing the reliance on third-country providers.
This follows the launch of the EU’s InvestAI initiative, which pledged €200 billion towards developing and deploying its own AI models. Likewise, an announcement is
expected
next month detailing the bloc’s plans to triple its data center capacity within seven years, as part of a broader AI action plan.
The foundation of Europe’s AI ambitions
Discussions around AI sovereignty tend to focus on where data is stored and processed, largely ignoring the path it takes to get there.
Data labelling and annotation, training models, running inference, handling API calls, monitoring and logging, backing up, and much more all require a continuous flow of information, typically traveling across multiple countries and providers. Last year, AI-driven traffic nearly tripled, with traffic from AI agents and agentic browsers growing by close to
8,000%
year-over-year. Without a reliable network supporting it, that information struggles to get around, training cycles become longer, and inference slows.
Fortunately, Europe already benefits from a high-quality, highly interconnected network that will provide a reliable backbone for AI development. However, while the infrastructure is ready to support AI sovereignty, policy lags behind.
Serving a fragmented market
What allows emerging technologies to flourish in markets like the US and China is the ability to operate within a single, unified system. China’s push is being orchestrated largely by its government. Likewise, in the US, while regulation exists at the state level, AI’s rapid and unexpected growth has caused legislation to lag behind innovation. As a result, the market has remained largely unfragmented, allowing companies to scale rapidly without facing much complexity or challenge.
Plus, both countries heavily subsidise their cloud providers. Not directly, but through large government contracts that provide stable, long-term revenue. They’re essentially underwriting innovation, de-risking substantial investment in the infrastructure that enables hyperscalers to grow.
W
hile EU countries may be on the same page politically, the bloc is far from unfragmented. There are significant differences in data regulation, energy policies, and infrastructure investment that make it difficult for companies to operate across borders.
Despite the AI-readiness of Europe’s networks, fragmented policy will prevent it from being fully leveraged to achieve AI sovereignty. Success will require greater alignment on cross-border data governance to ensure companies aren’t constrained by how and where data can be accessed and processed, along with a more coordinated approach to digital infrastructure investment. Likewise, it needs a more streamlined regulatory framework for AI and cloud services, enabling organisations to deploy and scale across the bloc without inconsistent rules and requirements impacting economic viability.
Not so sovereign technology
Currently, the global internet ecosystem remains heavily dependent on US-based infrastructure and platforms. Even when data originates and terminates in Europe, traffic flows and services are often shaped by networks, cloud infrastructure, and interconnection hubs outside of the region.
American providers currently account for
83%
of the continent’s cloud infrastructure market, for instance. And while many now offer “sovereign” services that ensure data is stored and processed within EU borders, it still puts it in the hands of US-owned platforms, which causes concern.
Take the US Cloud Act, which insists US companies must comply with any demands for data that the US Government makes, regardless of where it’s stored. Complying with a request involving EU data stored in the EU would constitute a breach of GDPR, which begs the question: how does Europe plan to expand its AI capabilities
and
maintain digital sovereignty, while continuing to rely so heavily on US-controlled providers? You cannot be independent while building systems on top of infrastructure that is far from it.
That said, Europe is beginning to take steps to address the problem: The European Commission recently awarded four European providers a tender worth up to
€180M
over six years, enabling EU institutions to procure their sovereign cloud services. It’s a clear sign of intent that, as it seeks AI sovereignty, work is already underway to strengthen control over the underlying infrastructure that supports it.
Supporting Europe’s backbone
Europe is showing clear ambition in challenging the US-China dominance, and it certainly has the infrastructure to achieve it.
The continent boasts one of the world’s most impressive networks, with some of the highest high-speed connectivity rates globally. It’s already capable of supporting a European AI industry – provided the right policy is put in place to support it.
AI sovereignty is not achieved simply by backing AI companies in Europe, developing models on the continent, or producing its own chips. As it increases its investment, Europe must consider how wider issues – such as cross-border data governance and dependency on outsider providers, which divide the continent and invite external influence in – could undermine its push for independence.
