Tech Sovereignty: Can the EU drive a wedge into the US-dominated tech stack?

The European Commission is expected to publish its Tech Sovereignty Package on 27 May. The initiative aims to strengthen Europe’s position in semiconductors, cloud infrastructure and artificial intelligence. Together, these three technologies provide the core capacities required to process, store and exploit data – the foundation of the digital economy. These technologies are sometimes described as constituting the “tech stack”, where each technology forms a distinct layer.

The concept of “sovereignty” reflects the Commission’s ambition to strengthen Europe’s position in the tech stack. Europe is currently heavily dependent on US firms in all three layers, which generates significant security risks regarding US access restrictions, export controls and surveillance, among other things. In the long term, Europe’s weak position in the stack risks undermining competitiveness and growth. US firms’ scale advantages and technological lead make full EU sovereignty unrealistic. Instead, it should take a more targeted approach, based on a carefully calibrated policy mix that takes full account of the governance constraints inherent in its decentralized political system.

US firms dominate the stack and are pulling further ahead

The dominance of US firms is evident in all three layers targeted by the Tech Sovereignty Package. Europe has important firms in each layer, but these are smaller actors, niche players or firms focused on complementary segments such as telecom infrastructure. The main exception is the Netherlands-based ASML, which has a globally strong position in lithography machinery, a central part of chip production. Research indicates that European companies are often deeply integrated into US technological ecosystems.

Table 1 shows how US dominance is underpinned by enormous economies of scale that operate across the entire stack, with strong tendencies towards vertical integration and winner-takes-all dynamics. NVIDIA’s recent $1 trillion agreement to supply Amazon with advanced chips illustrates the immense scale of these ecosystems.

These scale advantages, which are embedded in a highly dynamic US innovation ecosystem, make it very unlikely that the EU will be able to build an autonomous and globally competitive stack. Its strategy should instead focus on building strategic pockets of autonomy and leadership based on existing strengths.

What would an effective EU policy mix look like?

The challenge is not simply to deploy public resources, but to support market forces to generate a “demand-supply loop”, as described in a recent Dutch non-paper. This will require careful calibration of investment, regulation and international partnerships.

Investment

Inspired by Mario Draghi’s calls for large-scale industrial investment, the Commission has proposed allocating approximately €70 billion from the new European Competitiveness Fund to support digital leadership. Member states are likely to contribute additional investments of similar magnitude through Important Projects of Common European Interest (IPCEIs) focused on AI, compute infrastructure and semiconductors.

However, prioritization will raise tough questions at each layer of the stack, notably:

• how best to support Europe’s various strongholds in industrial AI, such as the automobile and robotics sectors.
• how to calibrate support across the different infrastructures needed for cloud and computing (candidates include EuroHPC supercomputers, quantum computing, and telecoms and electricity infrastructure), while ensuring that end-users benefit from these investments.
• how to weigh boosting Europe’s indispensability in specific parts of the semiconductor value chain – particularly manufacturing equipment – against enhancing resilience more broadly and moving towards the EU target of a 20% share of the global semiconductor market.

Regulation

The EU has extensive regulatory frameworks governing data, AI, digital platforms and cybersecurity (e.g. GDPR, the AI Act, the Digital Services Act, NIS2 and the Cyber Resilience Act). The EU is also seeking to facilitate data sharing within the internal market through legislation such as the Data Act, the Digital Networks Act and sector-specific initiatives such as European Common Data Spaces.

While there is currently strong pressure to simplify existing rules, there are also calls to use EU regulation strategically to boost market demand for sovereign offers. If the EU is to move further in that direction, options include:

• creating preferences for EU-based providers in public procurement and sensitive sectors, potentially in combination with restrictions on foreign cloud suppliers.
• assertive promotion of European standards in areas such as health data spaces and AI sandboxes, while ensuring close collaboration with key international partners.
• strictly enforced localization requirements or infrastructure fees imposed on large cloud providers.

International partnerships

Investment and regulation must be complemented by international partnerships. The EU will need to navigate growing US pressure – particularly concerning China – while seeking partnerships that reduce excessive dependence and enhance EU firms’ access to international markets.

Possible avenues include:

• expanding digital partnerships and trusted trade frameworks with countries such as Canada and Japan to promote open and rules-based digital governance.
• developing investment partnerships similar to the semiconductor factory in Dresden, in which Taiwan-based TSMC plays a central role.
• securing access to critical minerals through bilateral agreements and strategic diversification away from China, including by joining US-led Pax Silica.
• supporting open-source standards and interoperable systems to reduce dependence on proprietary systems.

What policies can the EU realistically implement?

Any strategy must ultimately take account of the realities of EU governance. Research suggests that EU industrial policy is often shaped by politicized bargaining between member states and EU legislators, at the expense of efficiency and expertise-based prioritization. Some researchers argue that new expert-based governance capacities are emerging, for instance for deciding on IPCEIs, but technocratic coordination alone will not suffice.

In the end, elected politicians at EU and national level will need to decide how much collective technological sovereignty Europe is willing to pay for. A key political consideration in that context will be the EU’s willingness and capacity to manage US pushback and retaliation in pursuit of greater technological autonomy.