Europe’s answer to OpenAI launched last year. Now it’s worth €6bn

Arthur Mensch is a celebrity in Paris. The entrepreneur is the co-founder and chief executive of Mistral, Europe’s answer to OpenAI, which has shot to fame not only in his home country but also in Silicon Valley.
Named after the strong, cold wind that blows from southern France into the Mediterranean, Mistral is hailed by Mensch, 32, a former academic, as “a French wind and a wind of change”. In case there’s any doubt about its origins, the name Mistral was chosen, too, because it included the letters IA, in that order, which stand for intelligence artificielle, French for AI.
It was a wind of change that seemed to come from nowhere. The company was started only last year, founded by Mensch, Guillaume Lample and Timothée Lacroix, also in their early thirties. The former researchers from Google DeepMind and Meta met while at the École Polytechnique and École Normale Supérieure, two of France’s grandes écoles, which specialise in science and engineering. After developing AI for American companies, they set out to do the same in Europe.
Mistral secured funding from some of the best-known American investment firms, including Lightspeed Venture and Andreessen Horowitz, at astounding speed. Eric Schmidt, the former Google chief executive, put money in, as did famous French names such as Xavier Niel, the telecoms billionaire.
Paul Murphy, who was part of the team that brought Mistral to the attention of Lightspeed, is a partner at the fund’s London office and sits on the company’s board. “The founders are probably three of the top ten people from a research perspective in the field of AI right now,” he said “When we invested in the company, it was already off to the races.”
The business builds large-language models for generative AI, technology trained on vast swathes of data so that it can learn from examples to create new content, such as writing stories. It has a chatbot called Le Chat (pronounced as per the French for cat, rather than the English for talking) and a platform for developers called La Plateforme. In February Mistral AI announced a revenue-sharing partnership with Microsoft to distribute its models. It has similar deals with AWS and Databricks.
It celebrated its first birthday in June by announcing that it had raised further funding of €600 million from backers including General Catalyst, Nvidia, Salesforce, Samsung and IBM. This took the total raised by the business in less than a year to more than €1 billion, giving it an extraordinary valuation of almost €6 billion, propelling it into a leading position in AI but keeping it under its founders’ control. The speed at which it launched its first products took everyone in the industry by surprise.
There are only a handful of so-called frontier AI companies around the world. Most are in the United States, so sovereignty is highly prized. In the global race to control AI, Mistral’s European roots matter and may become even more important. On July 19, Yann LeCun, Meta’s chief AI scientist, wrote: “Meta will *not* release the multimodal versions of its AI products and models in the EU because of an unpredictable regulatory environment.”
Unlike OpenAI and Anthropic but akin to Meta, Mistral’s technology is “open source”, so the company shares its tools and software with researchers, developers and companies to refine and modify. This, Mensch argues, is an effective way to improve the product.
Murphy said it also gave Lightspeed the conviction to invest. “AI is very different to other forms of software and that there’s no way that the most confidential data is going to get sent over to a black box.”
Mistral AI focuses on complex use cases in finance, technology and the public sector. It recently hired Marjorie Janiewicz, a former chief revenue officer at Foursquare and a French technology ambassador in San Francisco, to run its American operations, as the company is juggling more than 1,000 sales leads from some of the world’s biggest names. Although the team remains very small in comparison with its peers, it is hiring carefully. Mistral has compiled a list of 50 key people in AI that it wants to hire and has managed already to bring some of them on board.
Many onlookers wonder: “Why France?” Mensch attributes the success of AI across the Channel to the country’s renowned computer science and maths educational system, along with a renaissance of its technology scene. Over the past decade, Paris has nurtured more and more technology businesses, championed by President Macron, who vowed to make France a “start-up nation”, setting his sights on challenging the dominance of the sector by America and China.
“Europe has been producing some of the best researchers and scientists and technology for decades, but never got any credit because they would just get exported to the US,” Murphy said. “Now you have some of the best companies in the world being built across Europe and in the UK by people from Europe and the UK, invested in by funds like us that have a base in the region. So the whole life cycle is finally working in a way that it has always worked in the US and Silicon Valley.”
Macron has stood firmly behind the company. When Europe was creating AI legislation, he fought against strict rules for a more pro-innovation approach. And political ties extend within the business. One of the company’s board members is Cédric O, 41, a former French digital minister who evangelises about the importance of fighting to accelerate European technology.
“Geopolitics and technology, it’s a war of movement, not of position,” he recently told Sifted, a media platform covering the start-up sector. “The last time France thought it had solved its problem by taking shelter behind the Maginot line, it didn’t end so well.”
• Secure jobs, big salaries, but we gave them up to restart in AI
Beyond the hype, there remain challenges. Not least because the political wind has changed. France’s uncertain electoral outcome has created instability, not to mention uncertainty over whether the pro-business landscape will be maintained. Then there are practical considerations. Mistral has warned of a lack of data centre and grid capacity in Europe to train its models.
“We are reaching the capacity,” Audrey Herblin-Stoop, its head of public affairs, said: “We need to build data centres and ensure there is enough electricity for the scale of AI development today.”
Although the company’s valuation is enormous and the backing significant, it is still less than its Silicon Valley competitors. OpenAI has received $13 billion from Microsoft, Anthropic almost $8 billion and Elon Musk’s xAI $6 billion, funds needed for the expensive computing power, infrastructure and talent that underpins them. Above everything, Mistral’s challenge remains managing the next stage of growth in a highly competitive and rapidly changing world.