Ziad Choueiri, you say that the current focus on the cloud is necessary but largely insufficient. Why is this paradigm shift crucial today?
Ziad Choueiri: The cloud is only the foundation. It is the basic infrastructure. We have spent years securing facilities and ensuring the nationality of hosting providers’ capital, which is fundamental. But the true sovereignty of tomorrow will be decided at the next level – control of algorithms and overall data governance. In the age of AI, the risk is no longer just data leakage, but loss of control over decision-making.
The focus must now include questions about who truly controls the computational models, the intrinsic quality of the data that feeds them, and, above all, the explainability of algorithmic choices. Sovereignty is also a matter of cultural and organisational autonomy – are we capable of understanding and modifying the systems that manage our daily lives?
What does controlling algorithms mean in practical terms for a state or a European organisation?
Ziad Choueiri: It is a matter of protecting fundamental rights. For a state, controlling algorithms means ensuring that they comply with local laws and values and do not become uncontrollable black boxes. We must demand the supremacy of humans over machines. This means that no decision – whether it concerns the calculation of social benefits or a risk assessment – should be made without a human being able to audit it, explain it and, if necessary, overturn it.
Without such control, we expose ourselves to major discriminatory biases or public security vulnerabilities. Legal responsibility must remain clear; if an algorithm makes a mistake, who is accountable? If we lose control of the code, we lose control of responsibility.
Open source is often cited as the ideal safeguard. Is that your view?
Ziad Choueiri: It is an essential lever for transparency, but it is not a miracle solution. Open source prevents vendor lock-in, but it requires strong internal expertise to be maintained and industrialised. At Sopra Steria, we advocate a Sovereignty by Design approach.
The idea is to use open standards and promote interoperability. A sovereign system that does not communicate with the rest of the world is a dead end. True freedom is reversibility within a predictable, non-prohibitive timeframe and cost, the ability to leave a provider or technology without paralysing the service delivered.
We often hear about data quality, so what role does data play in the quest for sovereign AI?
Ziad Choueiri: It is the crux of the matter. AI performs well only if it is fed with high-quality data. Today, the main obstacle is not technological, but organisational. Siloed structures prevent a 360-degree view of the public service user.
Without strong governance of reference data (Master Data Management), AI will produce incomplete or erroneous results. To achieve sovereignty, the state must first be master of its own information. This requires political will to break down these silos so that data can circulate in a fluid and secure manner.
What will European public infrastructure look like in 2030 if we succeed in this ambition?
Ziad Choueiri: Let us be ambitious! By 2030, we will have distributed European hyperscalers with computing power capable of competing with American and Chinese giants. We will have harmonised data on sovereign matters, such as health, customs, security, education and justice.
We will have a legal framework that protects citizens without stifling innovation. But above all, we will have European champions at every stage of the AI stack, from leading AI models (frontier models) to operating systems. Europe will no longer be merely a consumer market for technology, but a sovereign innovation hub – a Silicon Valley of digital trust.