Building With Mistral—Europe’s Next Great Generative AI Startup

Founded by key members of Meta’s and Alphabet’s AI research teams, Mistral AI has closed a seed round of over €105 million, led by Lightspeed.

Mistral AI co-founders: Guillaume Lample, Arthur Mensch, Timothée Lacroix

Mistral AI, a European start-up specializing in generative artificial intelligence, has just announced that it has closed a seed-stage financing round of over €105 million led by Lightspeed.

Founded by key members of Meta’s and Alphabet’s AI research teams, Mistral AI will provide businesses the foundational technology to automate existing processes and develop innovative new products with state-of-the-art large language models.

Meet the team

Arthur, co-founder and CEO of Mistral met with Guillaume Lample and Timothée Lacroix, the other two co-founders building the Mistral trio, as students when studying at École Polytechnique and École Normale Supérieure.

Thanks to our international footprint, we’ve had the opportunity to meet several talented founders in the artificial intelligence sector. These profiles are often leading researchers, but few of them have a vision beyond the technical field.

When we met Arthur, Guillaume, and Timothée, we met a trio of seasoned researchers but, above all, entrepreneurs with a common set of values defining a strong vision.

Their goal was to create a European, open source project with the ambition of becoming a major global player in Generative AI. They also recognised the need to surround themselves with the right partners to make this ambition a reality beyond the technical sphere. The caliber of talent in and around Mistral is something we’ve rarely seen before and we’re excited to see what this complementary team will achieve.

The trio are part of a new breed of founders who combine deep technical expertise and operating experience working in the largest labs. This enables them to be part of a group of select few globally who have both the technical understanding required to build out their own vision along with the hands-on experience of training and operating large language models at scale.

The founders also share a common trait: they’ve had the opportunity to build deep technical expertise in their fields while working on state of the art applications of Machine Learning technology in some of the largest AI labs across the world. During his time at DeepMind, Arthur was a lead contributor to the Retro, Flamingo, and Chinchilla projects, gaining valuable experience in optimizing large language models. Guillaume led the development of LLaMa large language models along with Timothée. LLaMa took the world by storm at the beginning of the year, spurring a wave of open source enthusiasm after the weights of the models were leaked.

The case for open-source LLMs

This wave of open-source enthusiasm is one that we have been keeping a close eye on. Thanks to investments in the likes of Stability, Grafana, Yugabyte, Elementor, and Mulesoft, we’re no strangers to the unique opportunities that a strong open-source motion can bring to the fore.

Europe has built category-defining businesses leveraging open-source and Mistral intends to continue building on the shoulders of open-source giants (to name a few Grafana, Aiven, Odoo, Snyk, Elastic, Gitlab, Suse, Canonical distributing Ubuntu, MySQL/MariaDB, Blender). This open-source ethos is all the more critical in the context of AI where some of the world’s most prominent research is being conducted in walled gardens.

Mistral’s ambition is to build a credible open-source alternative to some of the early Large Language Model players while contributing and giving back to the community. For enterprise customers, the value of open source should not be underestimated. By committing to open source, Mistral avoids fears of lock-in and black box models that could raise regulatory and business concerns.

Europe is now

We’re particularly excited about what such a large round of funding means for the ecosystem. Europe has always had a strong research footing where technical talent was able to conduct deep research. To name a few, Stable Diffusion emerged as part of the CompVis group from the Ludwig Maximilian University of Munich, Scikit-learn, one of the most used Python packages for Machine Learning was developed by the Parietal group at Inria. Arthur Mensch, the CEO and co-founder of Mistral coincidentally (or not) got his PhD there!

We believe that Europe has a decisive role to play in the AI field. As the politicians and European leaders recognise the importance of their role in this new emerging world order, it is imperative that Europe remains a fertile ground for innovation to uphold European values but also further contribute to European digital sovereignty.

Following our office launches in the UK, Germany, and France, we are furthering our commitment to the region, and seeking to partner with more European founders who share the same ambition levels and vision that Arthur, Guillaume and Timothée possess. More generally speaking, it feels like the talent and capital flywheels are finally spinning at a sufficient speed to enable ambitious projects like Mistral to emerge and play a role on the international scene.

Our investment in Mistral, and all our portfolio companies in Europe, are evidence of our firm conviction that a new generation of global players will emerge from this ecosystem. We’re proud to be leading the way.

Lightspeed Possibility grows the deeper you go. Serving bold builders of the future.