Still unknown to the general public two years ago, Arthur Mensch is now one of the most promising faces in European artificial intelligence. After a notable stint at Google Deepmind, Mensch, who graduated from the Ecole Normale Supérieure, co-founded Mistral AI in 2023. The idea? Europe can--and must--offer a credible alternative to the American AI giants.

Betting on open, powerful and free-to-use language models, Mistral quickly shook up the codes of the sector. Where others lock down access to their technology, Mensch and his associates advocate transparent, reusable AI that serves collective innovation. A radical choice, which has already won over investors--with a record fundraising round--but also many companies and governments looking for sovereign solutions.

His presence at Nexus Luxembourg 2025 is therefore more than just a signal: it’s a way of reminding us that the AI battle is not just being fought in San Francisco or Beijing, but also here, in Europe, at the heart of institutions, financial markets... and ideas. In a world in search of technological and ethical benchmarks, Mensch’s voice counts. And in Luxembourg, it will be particularly resonant.

In less than two years, the startup has raised more than €1bn (including €600m in June 2024), reaching a valuation of several billion. Mistral claims to be “the only European player to provide what we provide” on the world stage. At the World Economic Forum in Davos in January 2025, Mensch insisted that “Mistral is not for sale” and confirmed that the preferred route would be an eventual stock market flotation.

At the same time, Mistral is stepping up its commercial development: Mensch said it had tripled its business in 100 days, particularly in Europe and outside the United States. It’s also opening offices abroad: in January 2025, it announced the opening of an office in Singapore to expand in Asia, whilst doubling its growth efforts in the US and Europe.

From the outset, Mensch has made open source a pillar of Mistral’s strategy. He sees open source as an engine for collective innovation: at the Mobile World Congress in Barcelona in March 2025, he explained that “the strength of open source is that different companies build on each other.” In practical terms, Mistral makes its model weights and inference methods publicly available, whilst protecting certain industrial secrets.

As Mensch explained in an interview, “we share the weights, we share the inference, we share a large part of the methods... There are obviously some manufacturing secrets that we keep.” The aim of this approach is to speed up research (other teams’ models can be based on his work) whilst maintaining a competitive edge. Mensch remains convinced that by combining openness and its own expertise, Mistral will be able to build more effective models: at the same congress in Barcelona, for example, he announced that its future versions would outperform the latest research from rival startups such as Deepseek.

Strategic partnerships and international reach

Under Mensch’s leadership, Mistral has stepped up its high-profile alliances. In early 2025, Mistral signed a content partnership with the AFP news agency to enrich its chatbot “Le Chat” with news dispatches. Commenting on this agreement, Mensch explained that “improving the accuracy” of Le Chat’s responses was “a key step in the deployment of our technology,” and that this partnership offered “our customers a unique multicultural and multilingual alternative.”

Internationally, Mistral has forged closer ties with major technology players: in June 2025, it announced a strategic partnership with UAE group G42 to co-develop next-generation AI platforms. “G42 is a partner that shares our commitment to making powerful and open AI accessible to all,” said Mensch. “This alliance not only accelerates our trajectory, but also ensures that the benefits of AI extend beyond traditional technology hubs.”

Mistral is also expanding its footprint in the fields of security and defence. In February 2025, at the Paris Global AI Summit, the startup announced a collaboration with the European company Helsing to develop AI systems for defence. At the time, Mesnch said that “with our partnership with Helsing, Europe is ready to take an important step towards asserting its technological superiority in AI. The versatile Mistral AI models are essential for developing a new generation of defence systems, guaranteeing Europe's strategic advantage.”

In the same spirit, Mistral in March 2025 signed an agreement with Singapore’s ministry of defence (Mindef, DSTA and DSO) to adapt its LLMs for military planning, and at the end of May it joined a contract with HTX (Singapore’s Home Team) and Microsoft to develop LLM models for public safety. Highlighting this regional commitment, Mensch said, “We are delighted to be working with Home Team to bring our expertise in LLM training and deployment to key initiatives... This collaboration reflects our continued investment and commitment to the APAC region.”

Book your ticket for Nexus Luxembourg 2025.

This article was originally published in French.