AI startup Mistral has raised €600 million ($644 million), tripling its valuation to $6 billion. This latest funding round (led by General Catalyst and featuring investors like Lightspeed Venture Partners, Andreessen Horowitz, Nvidia, and IBM) will fuel the development of both proprietary and open-source AI models.
Founded in April 2023 by former Meta and DeepMind employees, Mistral AI has experienced rapid growth, becoming one of the world’s most valuable open-source AI companies. CEO Arthur Mensch emphasized that the new capital will enhance the company’s computing capacity, recruitment efforts, and global expansion, particularly in the United States.
Mistral’s dual focus on proprietary and open-source models has driven its success. The company’s open-source models (such as Mistral 7B, Mixtral 8x7B, and Mixtral 8x22B) are accessible to developers under the Apache 2.0 license. These models have demonstrated competitive performance against leading commercial AI models on various benchmarks.
Revenue streams for Mistral include enterprise plans, licensing proprietary models on a pay-per-token basis, and a forthcoming API that supports model fine-tuning through usage fees or subscriptions. Additionally, Mistral has secured a multi-year partnership with Microsoft’s Azure platform, which includes a $16.3 million investment and equity participation.
Mistral’s valuation surge places it among the top AI companies globally, alongside OpenAI, Anthropic, and Databricks. Big congrats to Mistral, but before we get ahead of ourselves, let’s walk down memory lane.
Here are a few “foundational” search engines whose names I found in the attic of antiquity: JumpStation (1993), W3Catalog (1993), ALIWEB (1993), WWW Worm (1993), Infoseek (1994), Lycos (1994), WebCrawler (1994), Yahoo! (1994), Magellan (1995), Alta Vista (1995), Excite (1995), MetaCrawler (1995), HotBot (1996), Dogpile (1996), Northern Light (1997), Yandex (1997), Ask Jeeves “ask.com” (1997), MSN Search “Bing” (1998), Google (1998), Go.com (1998), AllTheWeb (1999), Teoma (1999), and Baidu (2000), just to name a few.
Which raises the question: How many foundational AI models will survive into adulthood?
Author’s note: This is not a sponsored post. I am the author of this article and it expresses my own opinions. I am not, nor is my company, receiving compensation for it. This work was created with the assistance of various generative AI models.