Europe has spent the better part of a decade learning how to regulate the digital world. In 2026, Brussels is trying something harder: building more of it.
The European Commission’s Technological Sovereignty Package, presented on 3 June, bundles a Chips Act 2.0, a Cloud and AI Development Act, an Open Source Strategy and a plan for digitalisation and AI in energy. It is a notable change of posture. The European Union is no longer treating digital power mainly as a question of privacy, competition rules and platform accountability. It is trying to build capacity across the technological stack.
Commission President Ursula von der Leyen framed the problem in unusually physical terms: “We cannot afford to depend on others for the technologies that keep our hospitals running, our energy grids stable and our services secure.”
The scale of the dependency explains the new urgency. EU institutions put Europe’s reliance on non-EU countries at more than 80% for key digital products, services, infrastructure and intellectual property. In semiconductors, the EU accounts for only about 9% of the global market, less than half its 20% target for 2030. In cloud computing, the Commission says three non-EU hyperscalers control more than 70% of the European market, while the share held by EU providers fell from 29% in 2017 to 15% in 2022 and has since stagnated.
Those numbers have turned “digital sovereignty” from a conference slogan into industrial policy.
Europe is building the lower layers
The new strategy is broad because the dependency is broad. Europe needs chips to train and run models, data centres to host workloads, cloud services to connect businesses to compute, networks to move information, and software that can be inspected, maintained and adapted without asking permission from a single external vendor.
On 30 July, the EU launched a tender for up to seven AI Gigafactories. The plan offers up to €10 billion in EU and national public funding and is designed to unlock at least €20 billion in private investment. These facilities are supposed to combine advanced AI processors, software, cloud stacks, high-speed connectivity and energy-efficient data centres. They will sit alongside a network of AI factories already being developed across Europe.
The ambition reaches beyond terrestrial computing. IRIS², the EU’s secure connectivity system, is planned as a constellation of more than 290 satellites in low and medium Earth orbit, with service scheduled for 2030. Semiconductor pilot lines are moving into operation. The Cloud and AI Development Act is explicitly designed to address both a shortage of European compute capacity and dependence on foreign cloud providers.
In other words, Europe increasingly understands that sovereignty has a supply chain. If it cannot manufacture enough critical components, host workloads, train models or maintain secure connectivity under European jurisdiction, regulation alone cannot compensate.
Public opinion is not the obstacle. The Commission’s 2026 Digital Decade report says 85% of Europeans support investment in digital infrastructure developed in the EU, while 82% favour reducing dependence on non-EU suppliers.
There is, however, another layer of the stack that is much harder to build with a subsidy or a procurement rule. It is the layer closest to the user.
The missing layer is the social graph
Look at the European Commission’s own Digital Markets Act map of the platforms powerful enough to be called gatekeepers.
The 2026 DMA review lists TikTok, Facebook, Instagram and LinkedIn as gatekeeper social networks. WhatsApp and Messenger are the designated number-independent messaging services. YouTube is the gatekeeper video-sharing platform. The companies behind those services are ByteDance, Meta, Microsoft and Alphabet — not EU-headquartered groups.
This is not a claim that Europe has no successful consumer technology. It clearly does. Nor is it an argument that every digital product Europeans use should be European. The structural issue is narrower and more important: in the services where identity, relationships, audiences and daily attention accumulate, Europe does not currently operate a mass-market platform comparable in scale to the dominant social and messaging networks.
The audience numbers show what that means. Transparency reporting for the second half of 2025 put Instagram at about 288.7 million average monthly active recipients in the EU, Facebook at roughly 263 million and TikTok at around 178 million. Those figures use platform-specific methodologies and their audiences overlap, so they should not be added together. But each number on its own illustrates the scale a challenger would have to confront.
The behaviour beneath those platform numbers is even more revealing. In 2025, 82% of people aged 16 to 74 in the EU used instant messaging services and 67% used social networks. Among 16- to 29-year-olds, 89.3% used social networks.
That is the user layer: not a distant data centre, but the place where a family group chat lives, where a small business finds customers, where a musician builds an audience, where a teenager’s friendships are coordinated, where a political movement circulates a video and where a newsroom discovers what millions of people are discussing.
A semiconductor factory can be built before it has millions of customers. A social network cannot.
Why a better European messenger is not enough
This is where Europe’s industrial logic collides with network economics.
The EuroStack report, published in 2025 by a group led by Francesca Bria, Paul Timmers and Fausto Gernone, makes the problem brutally clear: “one can build many great instant messaging applications, but if WhatsApp is closed and all European users are already locked in there, all will fail”.
The point is not that WhatsApp is technically impossible to challenge. It is that the product a user values is not merely the app. It is everyone already inside it.
A new cloud provider can win a corporate contract by being cheaper, safer or more compliant. A new chip can be designed into the next generation of equipment. A new messaging service asks something more difficult: users must persuade their friends, relatives, colleagues, clients, schools and community groups to move with them. A social network asks creators to rebuild audiences and advertisers to follow. The switching cost is social, not only technical.
That distinction matters for policy. Europe has become highly sophisticated at identifying digital bottlenecks, but much of its sovereignty agenda still concentrates on infrastructure, enterprise technology and regulated access. Those are necessary foundations. They do not automatically create a consumer relationship.
The Digital Markets Act is beginning to attack precisely this problem. The Commission’s first review says alternative messaging services can now interoperate with WhatsApp and Messenger, and that new providers — including a European SME — have entered the market with new functions as a result. Interoperability is strategically important because it weakens the assumption that a user must abandon an existing network to try another service.
But opening a gate is not the same as creating traffic through it. Consumer adoption still depends on product quality, trust, branding, discovery, distribution and a reason to change behaviour.
Regulation can open markets; it cannot manufacture habit
The European model has often been caricatured as “America innovates, China scales, Europe regulates”. The 2026 policy shift is an explicit attempt to break that pattern. Europe wants to regulate and build.
Yet the user layer exposes the limit of treating industrial capacity as a complete answer. Europe could meet its semiconductor target, expand sovereign cloud capacity, deploy IRIS² and operate formidable AI infrastructure while still finding that its citizens open an American or Chinese platform first thing in the morning.
That would be a strange form of sovereignty: European compute underneath, foreign distribution on top.
The distinction also helps clarify what technological sovereignty is not. Henna Virkkunen, the Commission executive vice-president responsible for the portfolio, put it plainly in May: “Technological sovereignty doesn't mean that we are planning to work somehow in isolation in the future.” Her emphasis was on avoiding risky dependence on one company or one solution while continuing to work with partners.
That is a more realistic objective than digital autarky. The United States will remain a central European technology partner. Global software and hardware supply chains are too interconnected to nationalise by rhetoric. The problem emerges when dependence becomes so concentrated that Europe lacks a credible alternative, bargaining power or the ability to keep a critical function running under its own rules.
At the consumer layer, credible alternatives require a different toolbox.
The next industrial policy is distribution
Europe’s Open Source Strategy can help because shared code lowers the cost of building common infrastructure and reduces dependence on proprietary systems. Interoperability can help because it allows a new service to connect to an existing network instead of demanding a mass migration on day one. Public procurement can help create anchor customers for European software, cloud and communications tools.
But none of those measures by itself solves distribution.
If Europe wants a serious consumer layer, it needs to treat reach, social graphs and switching costs as strategic infrastructure. That means measuring success not only in fabs opened, compute installed or research grants allocated, but in users acquired and retained. It means making cross-platform communication a practical norm rather than an exceptional regulatory concession. It means giving European companies a route from public or enterprise adoption into the consumer market. And it means accepting that marketing, design, creator ecosystems and product culture are not superficial additions to “real” technology; they are how technology becomes power in everyday life.
There are signs that Brussels has recognised the gap. The DMA explicitly targets gatekeeper control over the route to end users. The new sovereignty package gives open source a place beside chips and cloud. The Commission’s cloud proposal speaks of credible European alternatives, not simply data localisation.
The unresolved question is whether those pieces will be connected.
Europe’s first generation of digital sovereignty was about rules. Its second is about industrial capacity. A third will have to be about the user.
The test will not be whether Europe can build a data centre, design a chip or fund an AI model. It will be whether a European can choose a European digital service without losing the people, customers and communities that make that service useful. Until that is possible at mass scale, the continent may own more of the technology beneath the internet while still renting the relationship with the person holding the phone.