Mobile World Congress

Mensch, CEO of Mistral Ai: 'European telcos invest in large data centres'

The co-founder of the French company, a European leader in the Ai field, urged companies not to use only US technologies whenever possible

by correspondent Luca Salvioli

Il CEO di Mistral AI, Arthur Mensch, a destra nella foto, interviene durante il Mobile World Congress in corso a Barcellona. REUTERS/Bruna Casas

3' min read

3' min read

BARCELONA_ Arthur Mensch, CEO of Mistral, the French company that has established itself as Europe's leading expertise in the field of generative artificial intelligence, was highly anticipated at the Mobile World Congress. Indeed, Mistral has a key position in an area where Europe has decided to invest, after the Paris summit, in order not to lag behind the United States and China.

In front of a specialised audience, operators in particular, he insisted on the need for European telcos to invest in building large data centres. 'There has definitely been a general wake-up call in Europe with the beginning of 2025,' he said, 'particularly to reduce dependence on the big US technology groups. There is a growing discussion and ambitions in Europe are growing. Europe can offer a lot, telcos have the opportunity to invest in big data centres. We are doing that, we are building our datacenter in France. The Ai revolution is also an opportunity to decentralise the cloud and have more players, not just the Americans'.

Loading...

Mensch has an engineering degree from École Polytechnique and Télécom Paris, then a PhD in computer science; he worked three years at DeepMind before co-founding Mistral. In his speech, he insisted on the need to collaborate with all European industry.

"We are working with telecoms, because Ai will have consequences for the characteristics of the networks, with very customised information. It will change the architecture, we are working on this with Orange," he stressed.

He called generative Ai a 'very horizontal' technology, which at '90 per cent is the same as a product for both consumers and companies', but what changes is the information used by the models, the data. The collaboration of European companies in the adoption of these agents is important to realise the specifications, which for finance are different from HR, to give an example.

Positive words for DeepSeek, the Chinese Ai that proved that it is possible to develop Ai inexpensively. "It is an efficient model and one that shows how open source in this context is starting to prove successful".

Mistral's CEO did not lash out at the Ai Act, rather he downplayed the risk that it hampers innovation, which is often complained about by companies in the sector: 'Actually, it is not our biggest problem. Rather, it is the fragmentation of the market that is the problem,' referring to the 27 member states, with all their differences, procedures and delays. 'Consolidation could be a good solution'.

Keeping a rather pragmatic and optimistic approach, he added that 'in the last two months we see more ambition on the part of European companies. There are opportunities for manufacturing, pharmaceuticals and automotive, adopting European technology solutions. But we have to be pragmatic: for some key infrastructures there is no alternative to American vendors'.

Ai 'raises an issue of sovereignty. If we want them to complement our work, we have to work on European models. And then there are very sensitive areas of application, such as defence. We must have an autonomous Ai'. Extending his message to collaboration, he said: 'We are good at educating and training mathematical models. But on data and different interactions we need to have more experts and applications within companies'. And for the near future, Mistral's focus will be on creating increasingly specialised and personalised Ai models.

Copyright reserved ©
Loading...

Brand connect

Loading...

Newsletter

Notizie e approfondimenti sugli avvenimenti politici, economici e finanziari.

Iscriviti