Tech giants have begun investing in “sovereign AI” to reduce Europe’s dependence on the United States.

“Sovereign AI is a relatively new term that emerged about last year,” Chris Gou, Cisco’s head of public policy, told CNBC.

Currently, many of the largest large language models (LLMs), such as OpenAI’s ChatGPT and Anthropic’s Claude, use US-based data centers to store data and process queries via the cloud.

This worries European politicians and regulators, who believe that dependence on American technology harms competitiveness and technological sustainability, writes CNBC.

On the topic: Nvidia will introduce computers for humanoid robots in 2025

Marc Andreessen: Artificial intelligence is creating “thousands” of jobs

Sovereign AI has been discussed before in Europe. For example, the General Data Protection Regulation (GDPR) issued in the EU requires companies to properly handle user data while maintaining confidentiality. High-profile cases in the European Union have also sparked debate over whether data on European citizens can be safely transferred across borders.

In 2020, the European Court of Justice invalidated the EU-US data sharing agreement on the grounds that it did not provide the same level of protection guaranteed in the EU under the GDPR. As a result, many cloud services were forced to move their infrastructure where data was stored and processed.

Filippo Sanesi, head of marketing and operations at French cloud company OVHCloud, said the company is seeing strong demand for its infrastructure as customers “understand the value of storing data in Europe.”

With this data, you can create AI products and services. Sanesi explained that these services must be sovereign, controlled, implemented and developed by the local population for the local population or companies.

In Italy, Italia 9B, the first LLM trained with Italian language data, was launched this summer. The goal of the project is to store data in a specific jurisdiction and use information from residents of that region so that the output of the AI ​​system is more dependent on local languages, culture and history.

Nvidia’s head of enterprise sales for Europe, the Middle East and Africa, David Hogan, believes the main problem with artificial intelligence is that most large models were trained primarily with Western data.

On the topic: “The bubble is bursting a little”: Bloomberg learned about OpenAI and Google’s problems with new AI models

Yandex plans to invest in the development of artificial intelligence in Indonesia

In Denmark, where Nvidia has a large presence, authorities are concerned that vital services such as healthcare and telecommunications are being provided by artificial intelligence systems that do not “reflect” Danish culture and values, Hogan said.

On November 13, Denmark published guidance on the use of artificial intelligence under the new EU AI law. Other EU countries are expected to follow the rules.

Earlier this week, Berlin-based search engine Ecosia and its Parisian counterpart Qwant announced a joint project to develop a European index from scratch to provide better results in French and German.

Bruno Zerbib, chief technology officer of French telecommunications operator Orange, also talks about the demand for sovereign AI. It is in talks with several large AI modeling companies to create a sovereign AI model that takes language and culture into account.