The race for masteryartificial intelligence takes a new turn with the emergence of what we callSovereign AI. This approach seeks to develop AI models that rely on local infrastructures, thereby addressing specific economic and cultural concerns.
Sovereign AI: a response to the challenges of competitiveness and technological resilience
The objective is to strengthen the competitiveness by focusing more on the local infrastructure. According to Chris Gow, public policy manager at Cisco in Brussels, the term is relatively new, but he reiterates that it is essential to address the growing concerns of European politicians and regulators about dependence on American technology.
Indeed, large language models developed by companies like OpenAI and Anthropic primarily use U.S.-based data centers to store data and process queries via the cloud. This dependence raises concerns about Europe's competitiveness and technological resilience. The general regulations on data protection (GDPR) of the European Union requires, for example, that companies manage user data in a secure and privacy-friendly manner.
The legal and policy implications of data localization
In 2020, the Court of Justice of the European Union invalidated a data sharing framework between the EU and the United States, finding that this pact did not offer the same level of protection as that guaranteed in the Union . This policy development has led to increased pressure for the localization of cloud infrastructures, where data is stored and processed locally, in accordance with European legislation.
Filippo Sanesi, global director of marketing and operations at OVHCloud, confirms that the French cloud service provider sees increasing demand for its infrastructures located in Europe. Businesses are beginning to understand the importance of having their data under a specific jurisdiction, making the development of locally controlled AI products and services possible.
Landmark initiatives in sovereign AI
Several countries in Europe have already launched ambitious projects to promote sovereign AI. Denmark, for example, unveiled a powerful supercomputer funded by revenue from popular weight loss drugs. This supercomputer aims to support local pharmaceutical and biotechnology research as well as other initiatives where AI can be a valuable tool.
In Italy, an “AI factory” anchored by a supercomputer was unveiled earlier this year. This project includes developing an Italian language model for government workers. In the UAE, a generative model of autonomous AI has been developed while India has committed $1.2 billion to initiatives including an AI supercomputer equipped with tens of thousands of chips.
The cultural and identity impact of sovereign AI
The adoption of sovereign AI is not only about technical and economic aspects; it also extends to cultural and identity issues. As Nvidia CEO Jensen Huang points out, this approach “encodes” a country’s culture, history and collective intelligence, making AI a cornerstone of modern economies.
This initiative aims to ensure that the results produced by AI systems are rooted in local languages, cultures and histories. This becomes crucial in vital sectors such as healthcare and telecommunications, where cultural specificities play an important role in the acceptance and effectiveness of technologies.
Challenges and Considerations for Sovereign AI
One of the main challenges lies in the high costs associated with the training of large linguistic models. It takes more than a billion dollars for the necessary materials, energy and qualified personnel. However, the desire to invest in these resources is driven by the immense potential of AI and the need to maintain a competitive advantage at the national level.
Moreover, some experts point out that sufficient computing power is essential, like electricity, for the modern world. Therefore, it is essential that this resource be accessible to all in order to avoid further widening the technological gap between nations.