Let’s take a look at the screenshot of the company’s official website:
Yes, you read that right, plus the company name, there are only 4 lines.
AI is moving fast: A four-week-old startup has raised $113 million in seed funding and will compete with OpenAI in building, training and applying large-scale language models and generative artificial intelligence.
Paris-based Mistral AI, co-founded by alumni from Google DeepMind and Meta, will focus on open-source solutions and target enterprises, creating what CEO Arthur Mensch considers the biggest challenge in the field right now: "Making AI useful" . The company plans to release the first text-based generative AI models in 2024.
Mistral AI noted that French investment bank Bpifrance and former Google CEO Eric Schmidt are also shareholders. The 105 million euro ($113 million at today's exchange rate) round values Mistral AI at 240 million euros ($260 million), sources close to the company confirmed. It's worth noting that the same numbers were circulating in the French media a month ago when speculation started to swirl about the company.
Mensch (center), his co-founders Timothee Lacroix (CTO, left) and Guillaume Lample (Chief Science Officer, right), all in their early thirties, met in school when they were both researching artificial intelligence field.
Unlike most companies, Mistral AI has no customers, just a small team of founders, and no actual product. However, their ambitious goal is to provide the best open-source generative AI models for the future, challenging the industry dominance of the $100-billion companies. The promise of a future in which power and profits are no longer entirely in the hands of a few companies has attracted massive investment.
To understand the importance of Mistral AI's valuation, we need to have a deep understanding of artificial intelligence and its financial needs. Building cutting-edge generative models requires significant financial resources, especially if competing with giants like GPT-4 or Claude. The training process alone requires huge costs. For example, Meta's LLaMa model consumed $5 million in only 21 days of training. Bringing these models to mainstream use required careful curation of datasets, months of GPU usage, and well-paid professional engineers.
Additionally, during inference, the cost of a model generating output can quickly escalate to millions of dollars per month. For example, OpenAI is rumored to cost about $700,000 per day to run ChatGPT. The financial requirements for AI development are staggering, which partly explains Mistral AI's huge valuation.
However, there are other factors at play. Mistral AI is notable because its founding team consists of well-known researchers who previously led prominent language learning model teams at tech giants such as Meta and DeepMind. These individuals have unique expertise and experience, being one of the few capable of building such advanced models. Their presence adds credibility to the company and brings with it the potential for innovation.
Additionally, Mistral AI's commitment to utilizing public data aligns with European data privacy protection regulations, which also makes it more attractive to investors. In an era when data privacy and protection are important concerns, Mistral AI's commitment to only use public data shows their emphasis on risk aversion and compliance.
However, it's not just potential compliance that differentiates it from other companies. The unique positioning of Mistral AI is to solve the problem of open source models and build effective internal models for enterprises, thus playing an important role in the development process of artificial intelligence products. This unique positioning may have attracted investors' attention, since Mistral AI's viewpoint is clearly different from that of potential competitors such as OpenAI or Anthropic.
Finally, it is also worth considering the French government's strong interest in nurturing domestic leaders in cutting-edge fields such as artificial intelligence. French President Emmanuel Macron has stressed the urgency of developing "artificial intelligence champions" at home to strengthen France's position in the fast-growing technology sector. The massive funding support for Mistral AI may reflect this policy orientation, showing a strong combination of public-private partnerships aimed at nurturing innovative start-ups in France.
The meteoric rise of Mistral AI can be seen as a fascinating case study in the field of modern artificial intelligence. Investing tens of millions of dollars in a four-week-old company can be groundbreaking or a disastrous misjudgment, a lesson in the history of AI investing. However, Mistral AI's gathering of top talent has the potential to become Europe's first AI leader, while also sparking the notion that Europe could set new precedents in AI regulation, bringing home-grown companies like Mistral AI Competitive Advantage. Time will tell whether these investors are visionary prophets or victims of madness.