Mistral, a Wannabe OpenAI of Europe, Seeks $300 Million
istral, an artificial intelligence startup founded by former Meta Platforms and Alphabet researchers, plans to raise an additional $300 million from investors just four months after raising $113 million in a seed round led by Lightspeed Venture Partners, according to two people familiar with the discussions.
The round is expected to value the Paris-based startup, which is developing an open-source large language model and has framed itself as the “OpenAI of Europe,” at over $1 billion before the investment, according to another person. It’s unclear which VC firms Mistral has spoken to about investing. Andreessen Horowitz, one of the most active investors in generative AI, is currently seeking an investment in an open-source LLM developer, according to a person familiar with the matter.
THE TAKEAWAY
• Startup plans to raise capital at more than $1 billion valuation
• Ex-Meta and Alphabet researchers started company in May
• Firm is developing an open-sourced large language model
A representative for Mistral did not return a request for comment.
Mistral (pronounced Mis-TRAHL) is building a language model based only on publicly available data. It is open-sourcing its models, taking a different approach than OpenAI. Mistral is also developing its products in line with stricter European regulations, like the European Union’s AI Act, and has emphasized privacy and security, according to its pitch deck. That focus could help it compete with U.S. based rivals such as OpenAI for European corporate customers. Language-specific LLMs that capture the nuances of different cultures may also outperform OpenAI in certain markets.
Though Mistral isn’t yet generating revenue, the startup plans to also offer its models through an application programming interface, similar to OpenAI’s strategy, for a fee, according to its pitch deck. One startup founder has said using Mistral’s LLM halved their costs compared to OpenAI’s ChatGPT 3.5, The Information previously reported.
The five-month-old startup has already run into backlash from some researchers, who have complained that its first open-source model lacked the safety guardrails that other models include. For instance, users were able to ask Mistral’s model how to commit suicide and assault people, said a researcher on X. After researchers pointed out these problems in the days following the model’s release, Mistral co-founder and CEO Arthur Mensch released instructions on how to add guardrails to the model to avoid this.
The company’s financing, which is in the early stages, is yet another example of how much capital is required to train language models. OpenAI’s Sam Altman has privately said that his company may need to raise as much as $100 billion in the coming years, The Information previously reported. It's also indicative of the growing investor appetite for AI startups, which have raised billions so far this year.
Mistral, named after a northern winter wind, was founded by Mensch, a former DeepMind research scientist, as well as Timothée Lacroix and Guillaume Lample, who both left Meta in June. The startup’s seed funding in June valued it at $260 million, according to PitchBook.
In its seed pitch deck, it said it planned to dedicate 1% of its funding to a non-profit foundation focused on open-source community development.